train scripts history

.gitignore

@@ -10,3 +10,13 @@ poetry.toml
 **/log/
 *.spec
 dist/
+*steps.zip
+*.pkl
+best_model.zip
+*.csv
+*.npz
+*.xml
+*.json
+*.yaml
+*.iml
+*.TXT

command.md

@@ -0,0 +1,14 @@
+训练（默认）
+bash train.sh
+
+测试（实时+显示画面)
+GYM_CPU_MODE=test GYM_CPU_TEST_MODEL=scripts/gyms/logs/Walk_R0_005/best_model.zip GYM_CPU_TEST_FOLDER=scripts/gyms/logs/Walk_R0_005/ GYM_CPU_TEST_NO_RENDER=0 GYM_CPU_TEST_NO_REALTIME=0 bash train.sh
+
+测试（无画面、非实时）
+GYM_CPU_MODE=test GYM_CPU_TEST_NO_RENDER=1 GYM_CPU_TEST_NO_REALTIME=1 bash train.sh
+
+retrain（继续训练）
+GYM_CPU_MODE=train GYM_CPU_TRAIN_MODEL=scripts/gyms/logs/Walk_R0_005/best_model.zip bash train.sh
+
+retrain+改训练超参
+GYM_CPU_MODE=train GYM_CPU_TRAIN_MODEL=scripts/gyms/logs/Walk_R0_004/best_model.zip GYM_CPU_TRAIN_LR=2e-4 GYM_CPU_TRAIN_CLIP_RANGE=0.13 GYM_CPU_TRAIN_BATCH_SIZE=256 YM_CPU_TRAIN_GAMMA=0.95 GYM_CPU_TRAIN_ENT_COEF=0.05 GYM_CPU_TRAIN_EPOCHS=8 bash train.sh

communication/server.py

@@ -1,5 +1,6 @@
 import logging
 import socket
+import time
 from select import select
 from communication.world_parser import WorldParser
 
@@ -10,15 +11,27 @@ class Server:
     def __init__(self, host: str, port: int, world_parser: WorldParser):
         self.world_parser: WorldParser = world_parser
         self.__host: str = host
-        self.__port: str = port
+        self.__port: int = port
-        self.__socket: socket = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
+        self.__socket: socket.socket = self._create_socket()
-        self.__socket.setsockopt(socket.IPPROTO_TCP, socket.TCP_NODELAY, 1)
         self.__send_buff = []
         self.__rcv_buffer_size = 1024
         self.__rcv_buffer = bytearray(self.__rcv_buffer_size)
 
+    def _create_socket(self) -> socket.socket:
+        sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
+        sock.setsockopt(socket.IPPROTO_TCP, socket.TCP_NODELAY, 1)
+        return sock
+
     def connect(self) -> None:
         logger.info("Connecting to server at %s:%d...", self.__host, self.__port)
+
+        # Always reconnect with a fresh socket object.
+        try:
+            self.__socket.close()
+        except OSError:
+            pass
+        self.__socket = self._create_socket()
+
         while True:
             try:
                 self.__socket.connect((self.__host, self.__port))
@@ -27,12 +40,19 @@ class Server:
                 logger.error(
                     "Connection refused. Make sure the server is running and listening on {self.__host}:{self.__port}."
                 )
+                time.sleep(0.05)
 
         logger.info(f"Server connection established to {self.__host}:{self.__port}.")
 
     def shutdown(self) -> None:
-        self.__socket.close()
+        try:
             self.__socket.shutdown(socket.SHUT_RDWR)
+        except OSError:
+            pass
+        try:
+            self.__socket.close()
+        except OSError:
+            pass
 
     def send_immediate(self, msg: str) -> None:
         """

scripts/commons/Server.py

@@ -1,9 +1,10 @@
 import subprocess
 import os
+import time
 
 
 class Server():
-    def __init__(self, first_server_p, first_monitor_p, n_servers) -> None:
+    def __init__(self, first_server_p, first_monitor_p, n_servers, no_render=True, no_realtime=True) -> None:
         try:
             import psutil
             self.check_running_servers(psutil, first_server_p, first_monitor_p, n_servers)
@@ -17,21 +18,32 @@ class Server():
 
         # makes it easier to kill test servers without affecting train servers
         cmd = "rcssservermj"
+        render_arg = "--no-render" if no_render else ""
+        realtime_arg = "--no-realtime" if no_realtime else ""
         for i in range(n_servers):
             port = first_server_p + i
             mport = first_monitor_p + i
 
-            server_cmd = f"{cmd} --aport {port} --mport {mport} --no-render  --no-realtime"
+            server_cmd = f"{cmd} -c {port} -m {mport} {render_arg} {realtime_arg}".strip()
 
-            self.rcss_processes.append(
+            proc = subprocess.Popen(
-                subprocess.Popen(
                 server_cmd.split(),
                 stdout=subprocess.DEVNULL,
                 stderr=subprocess.STDOUT,
                 start_new_session=True
             )
+
+            # Avoid startup storm when launching many servers at once.
+            time.sleep(0.03)
+
+            rc = proc.poll()
+            if rc is not None:
+                raise RuntimeError(
+                    f"rcssservermj exited early (code={rc}) on server port {port}, monitor port {mport}"
                 )
 
+            self.rcss_processes.append(proc)
+
     def check_running_servers(self, psutil, first_server_p, first_monitor_p, n_servers):
         ''' Check if any server is running on chosen ports '''
         found = False

scripts/commons/Train_Base.py

@@ -171,8 +171,8 @@ class Train_Base():
             ep_reward += reward
             ep_length += 1
 
-            if enable_FPS_control:  # control simulation speed (using non blocking user input)
+            # if enable_FPS_control:  # control simulation speed (using non blocking user input)
-                self.control_fps(select.select([sys.stdin], [], [], 0)[0])
+            #     self.control_fps(select.select([sys.stdin], [], [], 0)[0])
 
             if done:
                 obs, _ = env.reset()

scripts/gyms/Walk.py

@@ -5,9 +5,11 @@ import time
 from time import sleep
 from random import random
 from random import uniform
+from itertools import count
 
 from stable_baselines3 import PPO
-from stable_baselines3.common.vec_env import SubprocVecEnv
+from stable_baselines3.common.monitor import Monitor
+from stable_baselines3.common.vec_env import SubprocVecEnv, DummyVecEnv
 
 import gymnasium as gym
 from gymnasium import spaces
@@ -50,6 +52,7 @@ class WalkEnv(gym.Env):
         self.waypoint_index = 0
         self.route_completed = False
         self.debug_every_n_steps = 5
+        self.enable_debug_joint_status = False
         self.calibrate_nominal_from_neutral = True
         self.auto_calibrate_train_sim_flip = True
         self.nominal_calibrated_once = False
@@ -57,7 +60,7 @@ class WalkEnv(gym.Env):
         self._target_hz = 0.0
         self._target_dt = 0.0
         self._last_sync_time = None
-        target_hz_env = 24
+        target_hz_env = 0
         if target_hz_env:
             try:
                 self._target_hz = float(target_hz_env)
@@ -80,8 +83,8 @@ class WalkEnv(gym.Env):
         action_dim = len(self.Player.robot.ROBOT_MOTORS)
         self.no_of_actions = action_dim
         self.action_space = spaces.Box(
-            low=-1.0,
+            low=-10.0,
-            high=1.0,
+            high=10.0,
             shape=(action_dim,),
             dtype=np.float32
         )
@@ -114,18 +117,17 @@ class WalkEnv(gym.Env):
                 0.0,
             ]
         )
-        self.reference_joint_nominal_position = self.joint_nominal_position.copy()
+        self.joint_nominal_position = np.zeros(self.no_of_actions)
-
         self.train_sim_flip = np.array(
             [
                 1.0,  # 0: Head_yaw (he1)
                 -1.0,  # 1: Head_pitch (he2)
                 1.0,  # 2: Left_Shoulder_Pitch (lae1)
                 -1.0,  # 3: Left_Shoulder_Roll (lae2)
-                1.0,  # 4: Left_Elbow_Pitch (lae3)
+                -1.0,  # 4: Left_Elbow_Pitch (lae3)
                 1.0,  # 5: Left_Elbow_Yaw (lae4)
                 -1.0,  # 6: Right_Shoulder_Pitch (rae1)
-                1.0,  # 7: Right_Shoulder_Roll (rae2)
+                -1.0,  # 7: Right_Shoulder_Roll (rae2)
                 1.0,  # 8: Right_Elbow_Pitch (rae3)
                 1.0,  # 9: Right_Elbow_Yaw (rae4)
                 1.0,  # 10: Waist (te1)
@@ -144,15 +146,54 @@ class WalkEnv(gym.Env):
             ]
         )
 
-        self.scaling_factor = 0.5
+        self.scaling_factor = 0.3
+        # self.scaling_factor = 1
+
+        # Encourage a minimum lateral stance so the policy avoids feet overlap.
+        self.min_stance_rad = 0.10
+
+        # Small reset perturbations for robustness training.
+        self.enable_reset_perturb = True
+        self.reset_beam_yaw_range_deg = 180  # randomize target direction fully to encourage learning a real walk instead of a fixed gait
+        self.reset_joint_noise_rad = 0.025
+        self.reset_perturb_steps = 4
+        self.reset_recover_steps = 8
 
         self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
         self.previous_pos = np.array([0.0, 0.0])  # Track previous position
         self.Player.server.connect()
         # sleep(2.0)  # Longer wait for connection to establish completely
         self.Player.server.send_immediate(
             f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
         )
+        self.start_time = time.time()
+
+    def _reconnect_server(self):
+        try:
+            self.Player.server.shutdown()
+        except Exception:
+            pass
+
+        self.Player.server.connect()
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+
+    def _safe_receive_world_update(self, retries=1):
+        last_exc = None
+        for attempt in range(retries + 1):
+            try:
+                self.Player.server.receive()
+                self.Player.world.update()
+                return
+            except (ConnectionResetError, OSError) as exc:
+                last_exc = exc
+                if attempt >= retries:
+                    raise
+                self._reconnect_server()
+        if last_exc is not None:
+            raise last_exc
 
     def debug_log(self, message):
         print(message)
@@ -163,45 +204,6 @@ class WalkEnv(gym.Env):
         except OSError:
             pass
 
-    def calibrate_train_sim_flip_from_neutral(self, neutral_joint_positions):
-        updated_flip = self.train_sim_flip.copy()
-        changed = []
-
-        for idx, (reference_value, observed_value) in enumerate(
-                zip(self.reference_joint_nominal_position, neutral_joint_positions)
-        ):
-            if idx >= 10:
-                continue
-            if abs(reference_value) < 0.15 or abs(observed_value) < 0.15:
-                continue
-
-            inferred_flip = 1.0 if np.sign(reference_value) == np.sign(observed_value) else -1.0
-            if updated_flip[idx] != inferred_flip:
-                changed.append((idx, updated_flip[idx], inferred_flip))
-                updated_flip[idx] = inferred_flip
-
-        self.train_sim_flip = updated_flip
-
-        if changed:
-            self.debug_log(
-                "[FlipDebug] "
-                f"changes={[(idx, old, new) for idx, old, new in changed]}"
-            )
-
-    def is_reliable_neutral_pose(self, neutral_joint_positions):
-        leg_positions = neutral_joint_positions[11:]
-        leg_norm = float(np.linalg.norm(leg_positions))
-        leg_max = float(np.max(np.abs(leg_positions)))
-        height = float(self.Player.world.global_position[2])
-
-        reliable = (
-                leg_norm > 0.8
-                and leg_max > 0.35
-                and 0.12 < height < 0.8
-        )
-
-        return reliable, leg_norm, leg_max, height
-
     def observe(self, init=False):
 
         """获取当前观测值"""
@@ -260,8 +262,7 @@ class WalkEnv(gym.Env):
 
     def sync(self):
         ''' Run a single simulation step '''
-        self.Player.server.receive()
+        self._safe_receive_world_update(retries=1)
-        self.Player.world.update()
         self.Player.robot.commit_motor_targets_pd()
         self.Player.server.send()
         if self._target_dt > 0.0:
@@ -299,6 +300,7 @@ class WalkEnv(gym.Env):
             f"err_norm={float(np.linalg.norm(joint_error)):.4f} "
             f"fallen={self.Player.world.global_position[2] < 0.3}"
         )
+        print(f"waist target={target_joint_positions[10]:.3f}, actual={actual_joint_positions[10]:.3f}")
 
     def reset(self, seed=None, options=None):
         '''
@@ -310,9 +312,9 @@ class WalkEnv(gym.Env):
         if seed is not None:
             np.random.seed(seed)
 
-        length1 = np.random.uniform(10, 20)  # randomize target distance
+        length1 = 2  # randomize target distance
-        length2 = np.random.uniform(10, 20)  # randomize target distance
+        length2 = np.random.uniform(0.6, 1)  # randomize target distance
-        length3 = np.random.uniform(10, 20)  # randomize target distance
+        length3 = np.random.uniform(0.6, 1)  # randomize target distance
         angle2 = np.random.uniform(-30, 30)  # randomize initial orientation
         angle3 = np.random.uniform(-30, 30)  # randomize target direction
 
@@ -320,70 +322,72 @@ class WalkEnv(gym.Env):
         self.waypoint_index = 0
         self.route_completed = False
         self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
         self.previous_pos = np.array([0.0, 0.0])  # Initialize for first step
         self.walk_cycle_step = 0
 
         # 随机 beam 目标位置和朝向，增加训练多样性
         beam_x = (random() - 0.5) * 10
         beam_y = (random() - 0.5) * 10
+        beam_yaw = uniform(-self.reset_beam_yaw_range_deg, self.reset_beam_yaw_range_deg)
 
         for _ in range(5):
-            self.Player.server.receive()
+            self._safe_receive_world_update(retries=2)
-            self.Player.world.update()
             self.Player.robot.commit_motor_targets_pd()
-            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=0)
+            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=beam_yaw)
             self.Player.server.send()
 
         # 执行 Neutral 技能直到完成，给机器人足够时间在 beam 位置稳定站立
         finished_count = 0
-        for _ in range(10):
+        for _ in range(50):
             finished = self.Player.skills_manager.execute("Neutral")
             self.sync()
             if finished:
                 finished_count += 1
-                if finished_count >= 3:  # 假设需要连续3次完成才算成功
+                if finished_count >= 20:  # 假设需要连续20次完成才算成功
                     break
 
-        # neutral_joint_positions = np.deg2rad(
+        if self.enable_reset_perturb and self.reset_joint_noise_rad > 0.0:
-        #     [self.Player.robot.motor_positions[motor] for motor in self.Player.robot.ROBOT_MOTORS]
+            perturb_action = np.zeros(self.no_of_actions, dtype=np.float32)
-        # )
+            # Perturb waist + lower body only (10:), keep head/arms stable.
-        # reliable_neutral, neutral_leg_norm, neutral_leg_max, neutral_height = self.is_reliable_neutral_pose(neutral_joint_positions)
+            perturb_action[10:] = np.random.uniform(
+                -self.reset_joint_noise_rad,
+                self.reset_joint_noise_rad,
+                size=(self.no_of_actions - 10,)
+            )
 
-        # if self.auto_calibrate_train_sim_flip and reliable_neutral and not self.flip_calibrated_once:
+            for _ in range(self.reset_perturb_steps):
-        #     self.calibrate_train_sim_flip_from_neutral(neutral_joint_positions)
+                target_joint_positions = (self.joint_nominal_position + perturb_action) * self.train_sim_flip
-        #     self.flip_calibrated_once = True
+                for idx, target in enumerate(target_joint_positions):
-        # if self.calibrate_nominal_from_neutral and reliable_neutral and not self.nominal_calibrated_once:
+                    r.set_motor_target_position(
-        #     self.joint_nominal_position = neutral_joint_positions * self.train_sim_flip
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
-        #     self.nominal_calibrated_once = True
+                    )
-        # self.debug_log(
+                self.sync()
-        #     "[ResetDebug] "
-        #     f"neutral_pos={np.round(self.Player.world.global_position, 3).tolist()} "
-        #     f"shoulders={np.round(neutral_joint_positions[2:10], 3).tolist()} "
-        #     f"legs={np.round(neutral_joint_positions[11:], 3).tolist()} "
-        #     f"flip={self.train_sim_flip.tolist()} "
-        #     f"nominal_legs={np.round(self.joint_nominal_position[11:], 3).tolist()} "
-        #     f"calibrated_once={(self.flip_calibrated_once, self.nominal_calibrated_once)} "
-        #     f"reliable_neutral={reliable_neutral} "
-        #     f"leg_norm={neutral_leg_norm:.3f} leg_max={neutral_leg_max:.3f} height={neutral_height:.3f}"
-        # )
 
-        # reset_action_noise = np.random.uniform(-0.015, 0.015, size=(len(self.Player.robot.ROBOT_MOTORS),))
+            for i in range(self.reset_recover_steps):
-        # self.target_joint_positions = (self.joint_nominal_position + reset_action_noise) * self.train_sim_flip
+                # Linearly fade perturbation to help policy start from near-neutral.
-
+                alpha = 1.0 - float(i + 1) / float(self.reset_recover_steps)
-        # for idx, target in enumerate(self.target_joint_positions):
+                target_joint_positions = (self.joint_nominal_position + alpha * perturb_action) * self.train_sim_flip
-        #     r.set_motor_target_position(
+                for idx, target in enumerate(target_joint_positions):
-        #         r.ROBOT_MOTORS[idx], target*180/math.pi, kp=25, kd=0.6
+                    r.set_motor_target_position(
-        #     )
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
 
         # memory variables
+        self.sync()  
         self.initial_position = np.array(self.Player.world.global_position[:2])
         self.previous_pos = self.initial_position.copy()  # Critical: set to actual position
         self.act = np.zeros(self.no_of_actions, np.float32)
-        point1 = self.initial_position + np.array([length1, 0])
+        # Build target in the robot's current forward direction instead of fixed global +x.
+        heading_deg = float(r.global_orientation_euler[2])
+        forward_offset = MathOps.rotate_2d_vec(np.array([length1, 0.0]), heading_deg, is_rad=False)
+        point1 = self.initial_position + forward_offset
         point2 = point1 + MathOps.rotate_2d_vec(np.array([length2, 0]), angle2, is_rad=False)
         point3 = point2 + MathOps.rotate_2d_vec(np.array([length3, 0]), angle3, is_rad=False)
-        self.point_list = [point1, point2, point3]
+        self.point_list = [point1]
         self.target_position = self.point_list[self.waypoint_index]
+        self.initial_height = self.Player.world.global_position[2]
 
         return self.observe(True), {}
 
@@ -391,83 +395,145 @@ class WalkEnv(gym.Env):
         return
 
     def compute_reward(self, previous_pos, current_pos, action):
-        velocity = current_pos - previous_pos
+        height = float(self.Player.world.global_position[2])
-        velocity_magnitude = np.linalg.norm(velocity)
+        robot = self.Player.robot
-        direction_to_target = self.target_position - current_pos
-        prev_direction_to_target = self.target_position - previous_pos
-        distance_to_target = np.linalg.norm(direction_to_target)
-        prev_distance_to_target = np.linalg.norm(prev_direction_to_target)
 
-        progress_reward = np.clip((prev_distance_to_target - distance_to_target) * 30.0, -2.0, 4.0)
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+        tilt_mag = float(np.linalg.norm(projected_gravity[:2]))
+        ang_vel = np.deg2rad(robot.gyroscope)
+        ang_vel_mag = float(np.linalg.norm(ang_vel))
 
-        velocity_in_m_per_sec = velocity_magnitude / 0.05
+        is_fallen = height < 0.55
-        speed_reward = np.clip(velocity_in_m_per_sec * 1.5, 0.0, 1.5)
+        if is_fallen:
+            # remain = max(0, 800 - self.step_counter)
+            # return -8.0 - 0.01 * remain
+            return -1.0
         
-        if velocity_magnitude > 1e-4 and distance_to_target > 1e-4:
-            directional_alignment = np.dot(velocity, direction_to_target) / (velocity_magnitude * distance_to_target)
-            directional_alignment = np.clip(directional_alignment, -1.0, 1.0)
-            direction_reward = max(0.0, directional_alignment)
-        else:
-            direction_reward = 0.0
 
-        alive_bonus = 0.05
 
-        height = self.Player.world.global_position[2]
+        # # 目标方向
-        if 0.45 <= height <= 1.2:
+        # to_target = self.target_position - current_pos
-            height_reward = 1.5
+        # dist_to_target = float(np.linalg.norm(to_target))
-        else:
+        # if dist_to_target < 0.5:
-            height_reward = -6.0
+        #     return 15.0
 
-        motionless_penalty = -1.5 if velocity_magnitude < 0.003 else 0.0
+        # forward_dir = to_target / dist_to_target if dist_to_target > 0.1 else np.array([1.0, 0.0])
+        # delta_pos = current_pos - previous_pos
+        # forward_step = float(np.dot(delta_pos, forward_dir))
+        # lateral_step = float(np.linalg.norm(delta_pos - forward_dir * forward_step))
 
-        waypoint_bonus = 0.0
+        # 奖励项
-        if distance_to_target < 0.5:
+        # progress_reward = 2 * forward_step
-            waypoint_bonus = 25.0
+        # lateral_penalty = -0.1 * lateral_step
-            if self.waypoint_index < len(self.point_list) - 1:
+        alive_bonus = 2.0
-                self.waypoint_index += 1
-                self.target_position = self.point_list[self.waypoint_index]
-            else:
-                waypoint_bonus = 100.0
-                self.route_completed = True
 
-        action_magnitude = np.linalg.norm(action[11:])
+        # action_penalty = -0.01 * float(np.linalg.norm(action))
-        action_penalty = -0.08 * action_magnitude
+        smoothness_penalty = -0.01 * float(np.linalg.norm(action - self.last_action_for_reward))
-        tilt_penalty = -0.2 * np.linalg.norm(self.Player.robot.gyroscope[:2]) / 100.0
 
-        return (
+        posture_penalty = -0.3 * (tilt_mag)
-                progress_reward
+        ang_vel_penalty = -0.02 * ang_vel_mag
-                + speed_reward
+
-                + direction_reward
+        # Use simulator joint readings in training frame to shape lateral stance.
-                + alive_bonus
+        joint_pos = np.deg2rad(
-                + height_reward
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
-                + motionless_penalty
+        ) * self.train_sim_flip
-                + waypoint_bonus
+        left_hip_roll = float(joint_pos[12])
-                + action_penalty
+        right_hip_roll = float(joint_pos[18])
-                + tilt_penalty
+        left_ankle_roll = float(joint_pos[16])
+        right_ankle_roll = float(joint_pos[22])
+
+        hip_spread = left_hip_roll - right_hip_roll
+        ankle_spread = left_ankle_roll - right_ankle_roll
+        stance_metric = 0.6 * abs(hip_spread) + 0.4 * abs(ankle_spread)
+
+        # Penalize narrow stance (feet too close) and scissoring (cross-leg pattern).
+        stance_collapse_penalty = -4.0 * max(0.0, self.min_stance_rad - stance_metric)
+        cross_leg_penalty = -1.2 * max(0.0, -(hip_spread * ankle_spread))
+
+        target_height = self.initial_height
+        height_error =  height - target_height
+        height_penalty = -0.5 * abs(height_error)  # 惩罚高度偏离，系数可调
+
+        # # 在 compute_reward 开头附近，添加高度变化率计算
+        # if not hasattr(self, 'last_height'):
+        #     self.last_height = height
+        #     self.last_height_time = self.step_counter  # 可选，用于时间间隔
+        # height_rate = height - self.last_height  # 正为上升，负为下降
+        # self.last_height = height
+
+        # 惩罚高度下降（负变化率）
+        # height_down_penalty = -5.0 * max(0, -height_rate)  # 系数可调，-height_rate 为正表示下降幅度
+
+        # # 在 compute_reward 中
+        # if self.step_counter > 50:
+        #     avg_prev_action = np.mean(self.prev_action_history, axis=0)
+        #     novelty = float(np.linalg.norm(action - avg_prev_action))
+        #     exploration_bonus = 0.05 * novelty
+        # else:
+        #     exploration_bonus = 0
+
+        # self.prev_action_history[self.history_idx] = action
+        # self.history_idx = (self.history_idx + 1) % 50
+
+
+        total = (
+            # progress_reward  + 
+                 alive_bonus + 
+                 # lateral_penalty +
+                 # action_penalty +
+                smoothness_penalty +
+                 posture_penalty
+                 + ang_vel_penalty
+                 + height_penalty
+                 + stance_collapse_penalty
+                 + cross_leg_penalty
+                #  + exploration_bonus
+                # + height_down_penalty
                  )
+        if time.time() - self.start_time >= 600:
+            self.start_time = time.time()
+            print(
+                #   f"progress_reward:{progress_reward:.4f}",
+                #   f"lateral_penalty:{lateral_penalty:.4f}",
+                #   f"action_penalty:{action_penalty:.4f}"s, 
+                  f"height_penalty:{height_penalty:.4f}",
+                  f"smoothness_penalty:{smoothness_penalty:.4f},",
+                  f"posture_penalty:{posture_penalty:.4f}",
+                  f"stance_collapse_penalty:{stance_collapse_penalty:.4f}",
+                  f"cross_leg_penalty:{cross_leg_penalty:.4f}",
+                #   f"ang_vel_penalty:{ang_vel_penalty:.4f}",
+                #   f"height_down_penalty:{height_down_penalty:.4f}",
+                #   f"exploration_bonus:{exploration_bonus:.4f}"
+            )
+
+        return total
+
+
 
     def step(self, action):
 
         r = self.Player.robot
-
         self.previous_action = action
 
         self.target_joint_positions = (
-                self.joint_nominal_position
+                # self.joint_nominal_position +
-                + self.scaling_factor * action
+                 self.scaling_factor * action
         )
         self.target_joint_positions *= self.train_sim_flip
 
         for idx, target in enumerate(self.target_joint_positions):
             r.set_motor_target_position(
-                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=40, kd=1.0
             )
 
+        self.previous_action = action
+
         self.sync()  # run simulation step
         self.step_counter += 1
 
-        # if self.step_counter % self.debug_every_n_steps == 0:
+        if self.enable_debug_joint_status and self.step_counter % self.debug_every_n_steps == 0:
-        #     self.debug_joint_status()
+            self.debug_joint_status()
 
         current_pos = np.array(self.Player.world.global_position[:2], dtype=np.float32)
 
@@ -476,9 +542,10 @@ class WalkEnv(gym.Env):
 
         # Update previous position
         self.previous_pos = current_pos.copy()
+        self.last_action_for_reward = action.copy()
 
         # Fall detection and penalty
-        is_fallen = self.Player.world.global_position[2] < 0.3
+        is_fallen = self.Player.world.global_position[2] < 0.55
 
         # terminal state: the robot is falling or timeout
         terminated = is_fallen or self.step_counter > 800 or self.route_completed
@@ -494,13 +561,14 @@ class Train(Train_Base):
     def train(self, args):
 
         # --------------------------------------- Learning parameters
-        n_envs = 8  # Reduced from 8 to decrease CPU/network pressure during init
+        n_envs = int(os.environ.get("GYM_CPU_N_ENVS", "20"))
         if n_envs < 1:
             raise ValueError("GYM_CPU_N_ENVS must be >= 1")
-        n_steps_per_env = 512  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        server_warmup_sec = float(os.environ.get("GYM_CPU_SERVER_WARMUP_SEC", "3.0"))
-        minibatch_size = 128  # should be a factor of (n_steps_per_env * n_envs)
+        n_steps_per_env = int(os.environ.get("GYM_CPU_TRAIN_STEPS_PER_ENV", "256"))  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        minibatch_size = int(os.environ.get("GYM_CPU_TRAIN_BATCH_SIZE", "512"))  # should be a factor of (n_steps_per_env * n_envs)
         total_steps = 30000000
-        learning_rate = 3e-4
+        learning_rate = float(os.environ.get("GYM_CPU_TRAIN_LR", "3e-4"))
         folder_name = f'Walk_R{self.robot_type}'
         model_path = f'./scripts/gyms/logs/{folder_name}/'
 
@@ -508,31 +576,38 @@ class Train(Train_Base):
         print(f"Using {n_envs} parallel environments")
 
         # --------------------------------------- Run algorithm
-        def init_env(i_env):
+        def init_env(i_env, monitor=False):
             def thunk():
-                return WalkEnv(self.ip, self.server_p + i_env)
+                env = WalkEnv(self.ip, self.server_p + i_env)
+                if monitor:
+                    env = Monitor(env)
+                return env
 
             return thunk
 
         server_log_dir = os.path.join(model_path, "server_logs")
         os.makedirs(server_log_dir, exist_ok=True)
-        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1)  # include 1 extra server for testing
+        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1, no_render=True, no_realtime=True)  # include 1 extra server for testing
 
         # Wait for servers to start
         print(f"Starting {n_envs + 1} rcssservermj servers...")
+        if server_warmup_sec > 0:
+            print(f"Waiting {server_warmup_sec:.1f}s for server warmup...")
+            sleep(server_warmup_sec)
         print("Servers started, creating environments...")
 
-        env = SubprocVecEnv([init_env(i) for i in range(n_envs)])
+        env = SubprocVecEnv([init_env(i, monitor=True) for i in range(n_envs)])
-        eval_env = SubprocVecEnv([init_env(n_envs)])
+        # Use single-process eval env to avoid extra subprocess fragility during callback evaluation.
+        eval_env = DummyVecEnv([init_env(n_envs, monitor=True)])
 
         try:
             # Custom policy network architecture
             policy_kwargs = dict(
                 net_arch=dict(
-                    pi=[256, 256, 128],  # Policy network: 3 layers
+                    pi=[512, 256, 128],  # Policy network: 3 layers
-                    vf=[256, 256, 128]  # Value network: 3 layers
+                    vf=[512, 256, 128]  # Value network: 3 layers
                 ),
-                activation_fn=__import__('torch.nn', fromlist=['ReLU']).ReLU,
+                activation_fn=__import__('torch.nn', fromlist=['ELU']).ELU,
             )
 
             if "model_file" in args:  # retrain
@@ -548,14 +623,17 @@ class Train(Train_Base):
                     learning_rate=learning_rate,
                     device="cpu",
                     policy_kwargs=policy_kwargs,
-                    ent_coef=0.01,  # Entropy coefficient for exploration
+                    ent_coef=float(os.environ.get("GYM_CPU_TRAIN_ENT_COEF", "0.05")),  # Entropy coefficient for exploration
-                    clip_range=0.2,  # PPO clipping parameter
+                    clip_range=float(os.environ.get("GYM_CPU_TRAIN_CLIP_RANGE", "0.2")),  # PPO clipping parameter
                     gae_lambda=0.95,  # GAE lambda
-                    gamma=0.99  # Discount factor
+                    gamma=float(os.environ.get("GYM_CPU_TRAIN_GAMMA", "0.95")), # Discount factor
+                    # target_kl=0.03,
+                    n_epochs=int(os.environ.get("GYM_CPU_TRAIN_EPOCHS", "5")),
+                    tensorboard_log=f"./scripts/gyms/logs/{folder_name}/tensorboard/"
                 )
 
             model_path = self.learn_model(model, total_steps, model_path, eval_env=eval_env,
-                                          eval_freq=n_steps_per_env * 10, save_freq=n_steps_per_env * 10,
+                                          eval_freq=n_steps_per_env * 20, save_freq=n_steps_per_env * 20, eval_eps=100,
                                           backup_env_file=__file__)
         except KeyboardInterrupt:
             sleep(1)  # wait for child processes
@@ -572,7 +650,16 @@ class Train(Train_Base):
         # Uses different server and monitor ports
         server_log_dir = os.path.join(args["folder_dir"], "server_logs")
         os.makedirs(server_log_dir, exist_ok=True)
-        server = Train_Server(self.server_p - 1, self.monitor_p, 1, log_dir=server_log_dir)
+        test_no_render = os.environ.get("GYM_CPU_TEST_NO_RENDER", "0") == "1"
+        test_no_realtime = os.environ.get("GYM_CPU_TEST_NO_REALTIME", "0") == "1"
+
+        server = Train_Server(
+            self.server_p - 1,
+            self.monitor_p,
+            1,
+            no_render=test_no_render,
+            no_realtime=test_no_realtime,
+        )
         env = WalkEnv(self.ip, self.server_p - 1)
         model = PPO.load(args["model_file"], env=env)
 
@@ -603,4 +690,16 @@ if __name__ == "__main__":
     )
 
     trainer = Train(script_args)
-    trainer.train({"model_file": "scripts/gyms/logs/Walk_R0_000/model_245760_steps.zip"})
+
+    run_mode = os.environ.get("GYM_CPU_MODE", "train").strip().lower()
+
+    if run_mode == "test":
+        test_model_file = os.environ.get("GYM_CPU_TEST_MODEL", "scripts/gyms/logs/Walk_R0_004/best_model.zip")
+        test_folder = os.environ.get("GYM_CPU_TEST_FOLDER", "scripts/gyms/logs/Walk_R0_004/")
+        trainer.test({"model_file": test_model_file, "folder_dir": test_folder})
+    else:
+        retrain_model = os.environ.get("GYM_CPU_TRAIN_MODEL", "").strip()
+        if retrain_model:
+            trainer.train({"model_file": retrain_model})
+        else:
+            trainer.train({})

scripts/gyms/logs/Walk_R0_000/Walk.py

@@ -0,0 +1,624 @@
+import os
+import numpy as np
+import math
+import time
+from time import sleep
+from random import random
+from random import uniform
+
+from stable_baselines3 import PPO
+from stable_baselines3.common.vec_env import SubprocVecEnv
+
+import gymnasium as gym
+from gymnasium import spaces
+
+from scripts.commons.Train_Base import Train_Base
+from scripts.commons.Server import Server as Train_Server
+
+from agent.base_agent import Base_Agent
+from utils.math_ops import MathOps
+
+from scipy.spatial.transform import Rotation as R
+
+'''
+Objective:
+Learn how to run forward using step primitive
+----------
+- class Basic_Run: implements an OpenAI custom gym
+- class Train:  implements algorithms to train a new model or test an existing model
+'''
+
+
+class WalkEnv(gym.Env):
+    def __init__(self, ip, server_p) -> None:
+
+        # Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
+        self.Player = player = Base_Agent(
+            team_name="Gym",
+            number=1,
+            host=ip,
+            port=server_p
+        )
+        self.robot_type = self.Player.robot
+        self.step_counter = 0  # to limit episode size
+        self.force_play_on = True
+
+        self.target_position = np.array([0.0, 0.0])  # target position in the x-y plane
+        self.initial_position = np.array([0.0, 0.0])  # initial position in the x-y plane
+        self.target_direction = 0.0  # target direction in the x-y plane (relative to the robot's orientation)
+        self.isfallen = False
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.debug_every_n_steps = 5
+        self.calibrate_nominal_from_neutral = True
+        self.auto_calibrate_train_sim_flip = True
+        self.nominal_calibrated_once = False
+        self.flip_calibrated_once = False
+        self._target_hz = 0.0
+        self._target_dt = 0.0
+        self._last_sync_time = None
+        target_hz_env = 0
+        if target_hz_env:
+            try:
+                self._target_hz = float(target_hz_env)
+            except ValueError:
+                self._target_hz = 0.0
+        if self._target_hz > 0.0:
+            self._target_dt = 1.0 / self._target_hz
+
+        # State space
+        # 原始观测大小: 78
+        obs_size = 78
+        self.obs = np.zeros(obs_size, np.float32)
+        self.observation_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(obs_size,),
+            dtype=np.float32
+        )
+
+        action_dim = len(self.Player.robot.ROBOT_MOTORS)
+        self.no_of_actions = action_dim
+        self.action_space = spaces.Box(
+            low=-1.0,
+            high=1.0,
+            shape=(action_dim,),
+            dtype=np.float32
+        )
+
+        # 中立姿态
+        self.joint_nominal_position = np.array(
+            [
+                0.0,
+                0.0,
+                0.0,
+                1.4,
+                0.0,
+                -0.4,
+                0.0,
+                -1.4,
+                0.0,
+                0.4,
+                0.0,
+                -0.4,
+                0.0,
+                0.0,
+                0.8,
+                -0.4,
+                0.0,
+                0.4,
+                0.0,
+                0.0,
+                -0.8,
+                0.4,
+                0.0,
+            ]
+        )
+        self.joint_nominal_position = np.zeros(self.no_of_actions)
+        self.train_sim_flip = np.array(
+            [
+                1.0,  # 0: Head_yaw (he1)
+                -1.0,  # 1: Head_pitch (he2)
+                1.0,  # 2: Left_Shoulder_Pitch (lae1)
+                -1.0,  # 3: Left_Shoulder_Roll (lae2)
+                -1.0,  # 4: Left_Elbow_Pitch (lae3)
+                1.0,  # 5: Left_Elbow_Yaw (lae4)
+                -1.0,  # 6: Right_Shoulder_Pitch (rae1)
+                -1.0,  # 7: Right_Shoulder_Roll (rae2)
+                1.0,  # 8: Right_Elbow_Pitch (rae3)
+                1.0,  # 9: Right_Elbow_Yaw (rae4)
+                1.0,  # 10: Waist (te1)
+                1.0,  # 11: Left_Hip_Pitch (lle1)
+                -1.0,  # 12: Left_Hip_Roll (lle2)
+                -1.0,  # 13: Left_Hip_Yaw (lle3)
+                1.0,  # 14: Left_Knee_Pitch (lle4)
+                1.0,  # 15: Left_Ankle_Pitch (lle5)
+                -1.0,  # 16: Left_Ankle_Roll (lle6)
+                -1.0,  # 17: Right_Hip_Pitch (rle1)
+                -1.0,  # 18: Right_Hip_Roll (rle2)
+                -1.0,  # 19: Right_Hip_Yaw (rle3)
+                -1.0,  # 20: Right_Knee_Pitch (rle4)
+                -1.0,  # 21: Right_Ankle_Pitch (rle5)
+                -1.0,  # 22: Right_Ankle_Roll (rle6)
+            ]
+        )
+
+        self.scaling_factor = 0.5
+        # self.scaling_factor = 1
+
+        # Small reset perturbations for robustness training.
+        self.enable_reset_perturb = False
+        self.reset_beam_yaw_range_deg = 180  # randomize target direction fully to encourage learning a real walk instead of a fixed gait
+        self.reset_joint_noise_rad = 0.035
+        self.reset_perturb_steps = 5
+        self.reset_recover_steps = 8
+
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Track previous position
+        self.Player.server.connect()
+        # sleep(2.0)  # Longer wait for connection to establish completely
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+        self.start_time = time.time()
+
+    def debug_log(self, message):
+        print(message)
+        try:
+            log_path = os.path.join(os.path.dirname(os.path.dirname(__file__)), "comm_debug.log")
+            with open(log_path, "a", encoding="utf-8") as f:
+                f.write(message + "\n")
+        except OSError:
+            pass
+
+    def observe(self, init=False):
+
+        """获取当前观测值"""
+        robot = self.Player.robot
+        world = self.Player.world
+
+        # Safety check: ensure data is available
+
+        # 计算目标速度
+        raw_target = self.target_position - world.global_position[:2]
+        velocity = MathOps.rotate_2d_vec(
+            raw_target,
+            -robot.global_orientation_euler[2],
+            is_rad=False
+        )
+
+        # 计算相对方向
+        rel_orientation = MathOps.vector_angle(velocity) * 0.3
+        rel_orientation = np.clip(rel_orientation, -0.25, 0.25)
+
+        velocity = np.concatenate([velocity, np.array([rel_orientation])])
+        velocity[0] = np.clip(velocity[0], -0.5, 0.5)
+        velocity[1] = np.clip(velocity[1], -0.25, 0.25)
+
+        # 关节状态
+        radian_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        radian_joint_speeds = np.deg2rad(
+            [robot.motor_speeds[motor] for motor in robot.ROBOT_MOTORS]
+        )
+
+        qpos_qvel_previous_action = np.concatenate([
+            (radian_joint_positions * self.train_sim_flip - self.joint_nominal_position) / 4.6,
+            radian_joint_speeds / 110.0 * self.train_sim_flip,
+            self.previous_action / 10.0,
+        ])
+
+        # 角速度
+        ang_vel = np.clip(np.deg2rad(robot.gyroscope) / 50.0, -1.0, 1.0)
+
+        # 投影的重力方向
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+
+        # 组合观测
+        observation = np.concatenate([
+            qpos_qvel_previous_action,
+            ang_vel,
+            velocity,
+            projected_gravity,
+        ])
+
+        observation = np.clip(observation, -10.0, 10.0)
+        return observation.astype(np.float32)
+
+    def sync(self):
+        ''' Run a single simulation step '''
+        self.Player.server.receive()
+        self.Player.world.update()
+        self.Player.robot.commit_motor_targets_pd()
+        self.Player.server.send()
+        if self._target_dt > 0.0:
+            now = time.time()
+            if self._last_sync_time is None:
+                self._last_sync_time = now
+                return
+            elapsed = now - self._last_sync_time
+            remaining = self._target_dt - elapsed
+            if remaining > 0.0:
+                time.sleep(remaining)
+                now = time.time()
+            self._last_sync_time = now
+
+    def debug_joint_status(self):
+        robot = self.Player.robot
+        actual_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        target_joint_positions = getattr(
+            self,
+            'target_joint_positions',
+            np.zeros(len(robot.ROBOT_MOTORS), dtype=np.float32)
+        )
+        joint_error = actual_joint_positions - target_joint_positions
+        leg_slice = slice(11, None)
+
+        self.debug_log(
+            "[WalkDebug] "
+            f"step={self.step_counter} "
+            f"pos={np.round(self.Player.world.global_position, 3).tolist()} "
+            f"target_xy={np.round(self.target_position, 3).tolist()} "
+            f"target_leg={np.round(target_joint_positions[leg_slice], 3).tolist()} "
+            f"actual_leg={np.round(actual_joint_positions[leg_slice], 3).tolist()} "
+            f"err_norm={float(np.linalg.norm(joint_error)):.4f} "
+            f"fallen={self.Player.world.global_position[2] < 0.3}"
+        )
+        print(f"waist target={target_joint_positions[10]:.3f}, actual={actual_joint_positions[10]:.3f}")
+
+    def reset(self, seed=None, options=None):
+        '''
+        Reset and stabilize the robot
+        Note: for some behaviors it would be better to reduce stabilization or add noise
+        '''
+        r = self.Player.robot
+        super().reset(seed=seed)
+        if seed is not None:
+            np.random.seed(seed)
+
+        length1 = 2  # randomize target distance
+        length2 = np.random.uniform(0.6, 1)  # randomize target distance
+        length3 = np.random.uniform(0.6, 1)  # randomize target distance
+        angle2 = np.random.uniform(-30, 30)  # randomize initial orientation
+        angle3 = np.random.uniform(-30, 30)  # randomize target direction
+
+        self.step_counter = 0
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Initialize for first step
+        self.walk_cycle_step = 0
+
+        # 随机 beam 目标位置和朝向，增加训练多样性
+        beam_x = (random() - 0.5) * 10
+        beam_y = (random() - 0.5) * 10
+        beam_yaw = uniform(-self.reset_beam_yaw_range_deg, self.reset_beam_yaw_range_deg)
+
+        for _ in range(5):
+            self.Player.server.receive()
+            self.Player.world.update()
+            self.Player.robot.commit_motor_targets_pd()
+            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=beam_yaw)
+            self.Player.server.send()
+
+        # 执行 Neutral 技能直到完成，给机器人足够时间在 beam 位置稳定站立
+        finished_count = 0
+        for _ in range(50):
+            finished = self.Player.skills_manager.execute("Neutral")
+            self.sync()
+            if finished:
+                finished_count += 1
+                if finished_count >= 20:  # 假设需要连续20次完成才算成功
+                    break
+
+        if self.enable_reset_perturb and self.reset_joint_noise_rad > 0.0:
+            perturb_action = np.zeros(self.no_of_actions, dtype=np.float32)
+            # Perturb waist + lower body only (10:), keep head/arms stable.
+            perturb_action[10:] = np.random.uniform(
+                -self.reset_joint_noise_rad,
+                self.reset_joint_noise_rad,
+                size=(self.no_of_actions - 10,)
+            )
+
+            for _ in range(self.reset_perturb_steps):
+                target_joint_positions = (self.joint_nominal_position + perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+            for i in range(self.reset_recover_steps):
+                # Linearly fade perturbation to help policy start from near-neutral.
+                alpha = 1.0 - float(i + 1) / float(self.reset_recover_steps)
+                target_joint_positions = (self.joint_nominal_position + alpha * perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+        # memory variables
+        self.initial_position = np.array(self.Player.world.global_position[:2])
+        self.previous_pos = self.initial_position.copy()  # Critical: set to actual position
+        self.act = np.zeros(self.no_of_actions, np.float32)
+        # Build target in the robot's current forward direction instead of fixed global +x.
+        heading_deg = float(r.global_orientation_euler[2])
+        forward_offset = MathOps.rotate_2d_vec(np.array([length1, 0.0]), heading_deg, is_rad=False)
+        point1 = self.initial_position + forward_offset
+        point2 = point1 + MathOps.rotate_2d_vec(np.array([length2, 0]), angle2, is_rad=False)
+        point3 = point2 + MathOps.rotate_2d_vec(np.array([length3, 0]), angle3, is_rad=False)
+        self.point_list = [point1]
+        self.target_position = self.point_list[self.waypoint_index]
+        self.initial_height = self.Player.world.global_position[2]
+
+        return self.observe(True), {}
+
+    def render(self, mode='human', close=False):
+        return
+
+    def compute_reward(self, previous_pos, current_pos, action):
+        height = float(self.Player.world.global_position[2])
+
+        orientation_quat_inv = R.from_quat(self.Player.robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+        tilt_mag = float(np.linalg.norm(projected_gravity[:2]))
+        ang_vel = np.deg2rad(self.Player.robot.gyroscope)
+        ang_vel_mag = float(np.linalg.norm(ang_vel))
+
+        # 摔倒检测（重要！）
+        if height < 0.3:
+            if time.time() - self.start_time > 1200:
+                self.start_time = time.time()
+                print("fall_penalty: -20")
+            return -20.0
+        
+
+
+        # # 目标方向
+        # to_target = self.target_position - current_pos
+        # dist_to_target = float(np.linalg.norm(to_target))
+        # if dist_to_target < 0.5:
+        #     return 15.0
+
+        # forward_dir = to_target / dist_to_target if dist_to_target > 0.1 else np.array([1.0, 0.0])
+        # delta_pos = current_pos - previous_pos
+        # forward_step = float(np.dot(delta_pos, forward_dir))
+        # lateral_step = float(np.linalg.norm(delta_pos - forward_dir * forward_step))
+
+        # 奖励项
+        # progress_reward = 2 * forward_step
+        # lateral_penalty = -0.1 * lateral_step
+        alive_bonus = 0.1
+
+        # action_penalty = -0.01 * float(np.linalg.norm(action))
+        smoothness_penalty = -0.05 * float(np.linalg.norm(action - self.last_action_for_reward))
+
+        posture_penalty = -1.0 * (tilt_mag)
+        # ang_vel_penalty = -0.05 * ang_vel_mag
+
+        target_height = self.initial_height
+        height_error =  height - target_height if abs(height - target_height) > 0.05 else 0.0
+        height_penalty = -2.0 * abs(height_error)  # 惩罚高度偏离，系数可调
+
+        # # 在 compute_reward 开头附近，添加高度变化率计算
+        # if not hasattr(self, 'last_height'):
+        #     self.last_height = height
+        #     self.last_height_time = self.step_counter  # 可选，用于时间间隔
+        # height_rate = height - self.last_height  # 正为上升，负为下降
+        # self.last_height = height
+
+        # 惩罚高度下降（负变化率）
+        # height_down_penalty = -5.0 * max(0, -height_rate)  # 系数可调，-height_rate 为正表示下降幅度
+
+        # # 在 compute_reward 中
+        # if self.step_counter > 50:
+        #     avg_prev_action = np.mean(self.prev_action_history, axis=0)
+        #     novelty = float(np.linalg.norm(action - avg_prev_action))
+        #     exploration_bonus = 0.05 * novelty
+        # else:
+        #     exploration_bonus = 0
+
+        # self.prev_action_history[self.history_idx] = action
+        # self.history_idx = (self.history_idx + 1) % 50
+
+
+        total = (
+            # progress_reward  + 
+                 alive_bonus + 
+                 # lateral_penalty +
+                 # action_penalty +
+                smoothness_penalty +
+                 posture_penalty
+                #  + ang_vel_penalty
+                 + height_penalty
+                #  + exploration_bonus
+                # + height_down_penalty
+                 )
+        if time.time() - self.start_time >= 1200:
+            self.start_time = time.time()
+            print(
+                #   f"progress_reward:{progress_reward:.4f}",
+                #   f"lateral_penalty:{lateral_penalty:.4f}",
+                #   f"action_penalty:{action_penalty:.4f}"s, 
+                  f"height_penalty:{height_penalty:.4f}",
+                  f"smoothness_penalty:{smoothness_penalty:.4f},",
+                  f"posture_penalty:{posture_penalty:.4f}",
+                #   f"ang_vel_penalty:{ang_vel_penalty:.4f}",
+                #   f"height_down_penalty:{height_down_penalty:.4f}",
+                #   f"exploration_bonus:{exploration_bonus:.4f}"
+            )
+
+        return total
+
+
+
+    def step(self, action):
+
+        r = self.Player.robot
+        self.previous_action = action
+
+        self.target_joint_positions = (
+                self.joint_nominal_position
+                + self.scaling_factor * action
+        )
+        self.target_joint_positions *= self.train_sim_flip
+
+        for idx, target in enumerate(self.target_joint_positions):
+            r.set_motor_target_position(
+                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+            )
+
+        self.previous_action = action
+
+        self.sync()  # run simulation step
+        self.step_counter += 1
+
+        # self.debug_joint_status()
+
+        current_pos = np.array(self.Player.world.global_position[:2], dtype=np.float32)
+
+        # Compute reward based on movement from previous step
+        reward = self.compute_reward(self.previous_pos, current_pos, action)
+
+        # Update previous position
+        self.previous_pos = current_pos.copy()
+        self.last_action_for_reward = action.copy()
+
+        # Fall detection and penalty
+        is_fallen = self.Player.world.global_position[2] < 0.3
+
+        # terminal state: the robot is falling or timeout
+        terminated = is_fallen or self.step_counter > 800 or self.route_completed
+        truncated = False
+
+        return self.observe(), reward, terminated, truncated, {}
+
+
+class Train(Train_Base):
+    def __init__(self, script) -> None:
+        super().__init__(script)
+
+    def train(self, args):
+
+        # --------------------------------------- Learning parameters
+        n_envs = 20  # Reduced from 8 to decrease CPU/network pressure during init
+        if n_envs < 1:
+            raise ValueError("GYM_CPU_N_ENVS must be >= 1")
+        n_steps_per_env = 1024  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        minibatch_size = 128  # should be a factor of (n_steps_per_env * n_envs)
+        total_steps = 30000000
+        learning_rate = 3e-4
+        folder_name = f'Walk_R{self.robot_type}'
+        model_path = f'./scripts/gyms/logs/{folder_name}/'
+
+        print(f"Model path: {model_path}")
+        print(f"Using {n_envs} parallel environments")
+
+        # --------------------------------------- Run algorithm
+        def init_env(i_env):
+            def thunk():
+                return WalkEnv(self.ip, self.server_p + i_env)
+
+            return thunk
+
+        server_log_dir = os.path.join(model_path, "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1)  # include 1 extra server for testing
+
+        # Wait for servers to start
+        print(f"Starting {n_envs + 1} rcssservermj servers...")
+        print("Servers started, creating environments...")
+
+        env = SubprocVecEnv([init_env(i) for i in range(n_envs)])
+        eval_env = SubprocVecEnv([init_env(n_envs)])
+
+        try:
+            # Custom policy network architecture
+            policy_kwargs = dict(
+                net_arch=dict(
+                    pi=[512, 256, 128],  # Policy network: 3 layers
+                    vf=[512, 256, 128]  # Value network: 3 layers
+                ),
+                activation_fn=__import__('torch.nn', fromlist=['ELU']).ELU,
+            )
+
+            if "model_file" in args:  # retrain
+                model = PPO.load(args["model_file"], env=env, device="cpu", n_envs=n_envs, n_steps=n_steps_per_env,
+                                 batch_size=minibatch_size, learning_rate=learning_rate)
+            else:  # train new model
+                model = PPO(
+                    "MlpPolicy",
+                    env=env,
+                    verbose=1,
+                    n_steps=n_steps_per_env,
+                    batch_size=minibatch_size,
+                    learning_rate=learning_rate,
+                    device="cpu",
+                    policy_kwargs=policy_kwargs,
+                    ent_coef=0.005,  # Entropy coefficient for exploration
+                    # clip_range=0.13,  # PPO clipping parameter
+                    gae_lambda=0.95,  # GAE lambda
+                    gamma=0.99 , # Discount factor
+                    # target_kl=0.03,
+                    # n_epochs=5
+                )
+
+            model_path = self.learn_model(model, total_steps, model_path, eval_env=eval_env,
+                                          eval_freq=n_steps_per_env * 10, save_freq=n_steps_per_env * 10,
+                                          backup_env_file=__file__)
+        except KeyboardInterrupt:
+            sleep(1)  # wait for child processes
+            print("\nctrl+c pressed, aborting...\n")
+            servers.kill()
+            return
+
+        env.close()
+        eval_env.close()
+        servers.kill()
+
+    def test(self, args):
+
+        # Uses different server and monitor ports
+        server_log_dir = os.path.join(args["folder_dir"], "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        server = Train_Server(self.server_p - 1, self.monitor_p, 1)
+        env = WalkEnv(self.ip, self.server_p - 1)
+        model = PPO.load(args["model_file"], env=env)
+
+        try:
+            self.export_model(args["model_file"], args["model_file"] + ".pkl",
+                              False)  # Export to pkl to create custom behavior
+            self.test_model(model, env, log_path=args["folder_dir"], model_path=args["folder_dir"])
+        except KeyboardInterrupt:
+            print()
+
+        env.close()
+        server.kill()
+
+
+if __name__ == "__main__":
+    from types import SimpleNamespace
+
+    # 创建默认参数
+    script_args = SimpleNamespace(
+        args=SimpleNamespace(
+            i='127.0.0.1',  # Server IP
+            p=3100,  # Server port
+            m=3200,  # Monitor port
+            r=0,  # Robot type
+            t='Gym',  # Team name
+            u=1  # Uniform number
+        )
+    )
+
+    trainer = Train(script_args)
+    trainer.train({})
+    # trainer.test({"model_file": "scripts/gyms/logs/Walk_R0_012/best_model.zip",
+                #   "folder_dir": "scripts/gyms/logs/Walk_R0_012/",})

scripts/gyms/logs/Walk_R0_001/Walk.py

@@ -0,0 +1,625 @@
+import os
+import numpy as np
+import math
+import time
+from time import sleep
+from random import random
+from random import uniform
+
+from stable_baselines3 import PPO
+from stable_baselines3.common.vec_env import SubprocVecEnv
+
+import gymnasium as gym
+from gymnasium import spaces
+
+from scripts.commons.Train_Base import Train_Base
+from scripts.commons.Server import Server as Train_Server
+
+from agent.base_agent import Base_Agent
+from utils.math_ops import MathOps
+
+from scipy.spatial.transform import Rotation as R
+
+'''
+Objective:
+Learn how to run forward using step primitive
+----------
+- class Basic_Run: implements an OpenAI custom gym
+- class Train:  implements algorithms to train a new model or test an existing model
+'''
+
+
+class WalkEnv(gym.Env):
+    def __init__(self, ip, server_p) -> None:
+
+        # Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
+        self.Player = player = Base_Agent(
+            team_name="Gym",
+            number=1,
+            host=ip,
+            port=server_p
+        )
+        self.robot_type = self.Player.robot
+        self.step_counter = 0  # to limit episode size
+        self.force_play_on = True
+
+        self.target_position = np.array([0.0, 0.0])  # target position in the x-y plane
+        self.initial_position = np.array([0.0, 0.0])  # initial position in the x-y plane
+        self.target_direction = 0.0  # target direction in the x-y plane (relative to the robot's orientation)
+        self.isfallen = False
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.debug_every_n_steps = 5
+        self.enable_debug_joint_status = False
+        self.calibrate_nominal_from_neutral = True
+        self.auto_calibrate_train_sim_flip = True
+        self.nominal_calibrated_once = False
+        self.flip_calibrated_once = False
+        self._target_hz = 0.0
+        self._target_dt = 0.0
+        self._last_sync_time = None
+        target_hz_env = 0
+        if target_hz_env:
+            try:
+                self._target_hz = float(target_hz_env)
+            except ValueError:
+                self._target_hz = 0.0
+        if self._target_hz > 0.0:
+            self._target_dt = 1.0 / self._target_hz
+
+        # State space
+        # 原始观测大小: 78
+        obs_size = 78
+        self.obs = np.zeros(obs_size, np.float32)
+        self.observation_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(obs_size,),
+            dtype=np.float32
+        )
+
+        action_dim = len(self.Player.robot.ROBOT_MOTORS)
+        self.no_of_actions = action_dim
+        self.action_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(action_dim,),
+            dtype=np.float32
+        )
+
+        # 中立姿态
+        self.joint_nominal_position = np.array(
+            [
+                0.0,
+                0.0,
+                0.0,
+                1.4,
+                0.0,
+                -0.4,
+                0.0,
+                -1.4,
+                0.0,
+                0.4,
+                0.0,
+                -0.4,
+                0.0,
+                0.0,
+                0.8,
+                -0.4,
+                0.0,
+                0.4,
+                0.0,
+                0.0,
+                -0.8,
+                0.4,
+                0.0,
+            ]
+        )
+        self.joint_nominal_position = np.zeros(self.no_of_actions)
+        self.train_sim_flip = np.array(
+            [
+                1.0,  # 0: Head_yaw (he1)
+                -1.0,  # 1: Head_pitch (he2)
+                1.0,  # 2: Left_Shoulder_Pitch (lae1)
+                -1.0,  # 3: Left_Shoulder_Roll (lae2)
+                -1.0,  # 4: Left_Elbow_Pitch (lae3)
+                1.0,  # 5: Left_Elbow_Yaw (lae4)
+                -1.0,  # 6: Right_Shoulder_Pitch (rae1)
+                -1.0,  # 7: Right_Shoulder_Roll (rae2)
+                1.0,  # 8: Right_Elbow_Pitch (rae3)
+                1.0,  # 9: Right_Elbow_Yaw (rae4)
+                1.0,  # 10: Waist (te1)
+                1.0,  # 11: Left_Hip_Pitch (lle1)
+                -1.0,  # 12: Left_Hip_Roll (lle2)
+                -1.0,  # 13: Left_Hip_Yaw (lle3)
+                1.0,  # 14: Left_Knee_Pitch (lle4)
+                1.0,  # 15: Left_Ankle_Pitch (lle5)
+                -1.0,  # 16: Left_Ankle_Roll (lle6)
+                -1.0,  # 17: Right_Hip_Pitch (rle1)
+                -1.0,  # 18: Right_Hip_Roll (rle2)
+                -1.0,  # 19: Right_Hip_Yaw (rle3)
+                -1.0,  # 20: Right_Knee_Pitch (rle4)
+                -1.0,  # 21: Right_Ankle_Pitch (rle5)
+                -1.0,  # 22: Right_Ankle_Roll (rle6)
+            ]
+        )
+
+        self.scaling_factor = 0.1
+        # self.scaling_factor = 1
+
+        # Small reset perturbations for robustness training.
+        self.enable_reset_perturb = False
+        self.reset_beam_yaw_range_deg = 180  # randomize target direction fully to encourage learning a real walk instead of a fixed gait
+        self.reset_joint_noise_rad = 0.035
+        self.reset_perturb_steps = 5
+        self.reset_recover_steps = 8
+
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Track previous position
+        self.Player.server.connect()
+        # sleep(2.0)  # Longer wait for connection to establish completely
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+        self.start_time = time.time()
+
+    def debug_log(self, message):
+        print(message)
+        try:
+            log_path = os.path.join(os.path.dirname(os.path.dirname(__file__)), "comm_debug.log")
+            with open(log_path, "a", encoding="utf-8") as f:
+                f.write(message + "\n")
+        except OSError:
+            pass
+
+    def observe(self, init=False):
+
+        """获取当前观测值"""
+        robot = self.Player.robot
+        world = self.Player.world
+
+        # Safety check: ensure data is available
+
+        # 计算目标速度
+        raw_target = self.target_position - world.global_position[:2]
+        velocity = MathOps.rotate_2d_vec(
+            raw_target,
+            -robot.global_orientation_euler[2],
+            is_rad=False
+        )
+
+        # 计算相对方向
+        rel_orientation = MathOps.vector_angle(velocity) * 0.3
+        rel_orientation = np.clip(rel_orientation, -0.25, 0.25)
+
+        velocity = np.concatenate([velocity, np.array([rel_orientation])])
+        velocity[0] = np.clip(velocity[0], -0.5, 0.5)
+        velocity[1] = np.clip(velocity[1], -0.25, 0.25)
+
+        # 关节状态
+        radian_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        radian_joint_speeds = np.deg2rad(
+            [robot.motor_speeds[motor] for motor in robot.ROBOT_MOTORS]
+        )
+
+        qpos_qvel_previous_action = np.concatenate([
+            (radian_joint_positions * self.train_sim_flip - self.joint_nominal_position) / 4.6,
+            radian_joint_speeds / 110.0 * self.train_sim_flip,
+            self.previous_action / 10.0,
+        ])
+
+        # 角速度
+        ang_vel = np.clip(np.deg2rad(robot.gyroscope) / 50.0, -1.0, 1.0)
+
+        # 投影的重力方向
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+
+        # 组合观测
+        observation = np.concatenate([
+            qpos_qvel_previous_action,
+            ang_vel,
+            velocity,
+            projected_gravity,
+        ])
+
+        observation = np.clip(observation, -10.0, 10.0)
+        return observation.astype(np.float32)
+
+    def sync(self):
+        ''' Run a single simulation step '''
+        self.Player.server.receive()
+        self.Player.world.update()
+        self.Player.robot.commit_motor_targets_pd()
+        self.Player.server.send()
+        if self._target_dt > 0.0:
+            now = time.time()
+            if self._last_sync_time is None:
+                self._last_sync_time = now
+                return
+            elapsed = now - self._last_sync_time
+            remaining = self._target_dt - elapsed
+            if remaining > 0.0:
+                time.sleep(remaining)
+                now = time.time()
+            self._last_sync_time = now
+
+    def debug_joint_status(self):
+        robot = self.Player.robot
+        actual_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        target_joint_positions = getattr(
+            self,
+            'target_joint_positions',
+            np.zeros(len(robot.ROBOT_MOTORS), dtype=np.float32)
+        )
+        joint_error = actual_joint_positions - target_joint_positions
+        leg_slice = slice(11, None)
+
+        self.debug_log(
+            "[WalkDebug] "
+            f"step={self.step_counter} "
+            f"pos={np.round(self.Player.world.global_position, 3).tolist()} "
+            f"target_xy={np.round(self.target_position, 3).tolist()} "
+            f"target_leg={np.round(target_joint_positions[leg_slice], 3).tolist()} "
+            f"actual_leg={np.round(actual_joint_positions[leg_slice], 3).tolist()} "
+            f"err_norm={float(np.linalg.norm(joint_error)):.4f} "
+            f"fallen={self.Player.world.global_position[2] < 0.3}"
+        )
+        print(f"waist target={target_joint_positions[10]:.3f}, actual={actual_joint_positions[10]:.3f}")
+
+    def reset(self, seed=None, options=None):
+        '''
+        Reset and stabilize the robot
+        Note: for some behaviors it would be better to reduce stabilization or add noise
+        '''
+        r = self.Player.robot
+        super().reset(seed=seed)
+        if seed is not None:
+            np.random.seed(seed)
+
+        length1 = 2  # randomize target distance
+        length2 = np.random.uniform(0.6, 1)  # randomize target distance
+        length3 = np.random.uniform(0.6, 1)  # randomize target distance
+        angle2 = np.random.uniform(-30, 30)  # randomize initial orientation
+        angle3 = np.random.uniform(-30, 30)  # randomize target direction
+
+        self.step_counter = 0
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Initialize for first step
+        self.walk_cycle_step = 0
+
+        # 随机 beam 目标位置和朝向，增加训练多样性
+        beam_x = (random() - 0.5) * 10
+        beam_y = (random() - 0.5) * 10
+        beam_yaw = uniform(-self.reset_beam_yaw_range_deg, self.reset_beam_yaw_range_deg)
+
+        for _ in range(5):
+            self.Player.server.receive()
+            self.Player.world.update()
+            self.Player.robot.commit_motor_targets_pd()
+            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=beam_yaw)
+            self.Player.server.send()
+
+        # 执行 Neutral 技能直到完成，给机器人足够时间在 beam 位置稳定站立
+        finished_count = 0
+        for _ in range(50):
+            finished = self.Player.skills_manager.execute("Neutral")
+            self.sync()
+            if finished:
+                finished_count += 1
+                if finished_count >= 20:  # 假设需要连续20次完成才算成功
+                    break
+
+        if self.enable_reset_perturb and self.reset_joint_noise_rad > 0.0:
+            perturb_action = np.zeros(self.no_of_actions, dtype=np.float32)
+            # Perturb waist + lower body only (10:), keep head/arms stable.
+            perturb_action[10:] = np.random.uniform(
+                -self.reset_joint_noise_rad,
+                self.reset_joint_noise_rad,
+                size=(self.no_of_actions - 10,)
+            )
+
+            for _ in range(self.reset_perturb_steps):
+                target_joint_positions = (self.joint_nominal_position + perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+            for i in range(self.reset_recover_steps):
+                # Linearly fade perturbation to help policy start from near-neutral.
+                alpha = 1.0 - float(i + 1) / float(self.reset_recover_steps)
+                target_joint_positions = (self.joint_nominal_position + alpha * perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+        # memory variables
+        self.sync()  
+        self.initial_position = np.array(self.Player.world.global_position[:2])
+        self.previous_pos = self.initial_position.copy()  # Critical: set to actual position
+        self.act = np.zeros(self.no_of_actions, np.float32)
+        # Build target in the robot's current forward direction instead of fixed global +x.
+        heading_deg = float(r.global_orientation_euler[2])
+        forward_offset = MathOps.rotate_2d_vec(np.array([length1, 0.0]), heading_deg, is_rad=False)
+        point1 = self.initial_position + forward_offset
+        point2 = point1 + MathOps.rotate_2d_vec(np.array([length2, 0]), angle2, is_rad=False)
+        point3 = point2 + MathOps.rotate_2d_vec(np.array([length3, 0]), angle3, is_rad=False)
+        self.point_list = [point1]
+        self.target_position = self.point_list[self.waypoint_index]
+        self.initial_height = self.Player.world.global_position[2]
+
+        return self.observe(True), {}
+
+    def render(self, mode='human', close=False):
+        return
+
+    def compute_reward(self, previous_pos, current_pos, action):
+        height = float(self.Player.world.global_position[2])
+
+        orientation_quat_inv = R.from_quat(self.Player.robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+        tilt_mag = float(np.linalg.norm(projected_gravity[:2]))
+        ang_vel = np.deg2rad(self.Player.robot.gyroscope)
+        ang_vel_mag = float(np.linalg.norm(ang_vel))
+
+        is_fallen = height < 0.3
+        if is_fallen:
+            remain = max(0, 800 - self.step_counter)
+            return -8.0 - 0.01 * remain
+        
+
+
+        # # 目标方向
+        # to_target = self.target_position - current_pos
+        # dist_to_target = float(np.linalg.norm(to_target))
+        # if dist_to_target < 0.5:
+        #     return 15.0
+
+        # forward_dir = to_target / dist_to_target if dist_to_target > 0.1 else np.array([1.0, 0.0])
+        # delta_pos = current_pos - previous_pos
+        # forward_step = float(np.dot(delta_pos, forward_dir))
+        # lateral_step = float(np.linalg.norm(delta_pos - forward_dir * forward_step))
+
+        # 奖励项
+        # progress_reward = 2 * forward_step
+        # lateral_penalty = -0.1 * lateral_step
+        alive_bonus = 0.3
+
+        # action_penalty = -0.01 * float(np.linalg.norm(action))
+        smoothness_penalty = -0.03 * float(np.linalg.norm(action - self.last_action_for_reward))
+
+        posture_penalty = -1.0 * (tilt_mag)
+        # ang_vel_penalty = -0.05 * ang_vel_mag
+
+        target_height = self.initial_height
+        height_error =  height - target_height if abs(height - target_height) > 0.05 else 0.0
+        height_penalty = -2.0 * abs(height_error)  # 惩罚高度偏离，系数可调
+
+        # # 在 compute_reward 开头附近，添加高度变化率计算
+        # if not hasattr(self, 'last_height'):
+        #     self.last_height = height
+        #     self.last_height_time = self.step_counter  # 可选，用于时间间隔
+        # height_rate = height - self.last_height  # 正为上升，负为下降
+        # self.last_height = height
+
+        # 惩罚高度下降（负变化率）
+        # height_down_penalty = -5.0 * max(0, -height_rate)  # 系数可调，-height_rate 为正表示下降幅度
+
+        # # 在 compute_reward 中
+        # if self.step_counter > 50:
+        #     avg_prev_action = np.mean(self.prev_action_history, axis=0)
+        #     novelty = float(np.linalg.norm(action - avg_prev_action))
+        #     exploration_bonus = 0.05 * novelty
+        # else:
+        #     exploration_bonus = 0
+
+        # self.prev_action_history[self.history_idx] = action
+        # self.history_idx = (self.history_idx + 1) % 50
+
+
+        total = (
+            # progress_reward  + 
+                 alive_bonus + 
+                 # lateral_penalty +
+                 # action_penalty +
+                smoothness_penalty +
+                 posture_penalty
+                #  + ang_vel_penalty
+                 + height_penalty
+                #  + exploration_bonus
+                # + height_down_penalty
+                 )
+        if time.time() - self.start_time >= 1200:
+            self.start_time = time.time()
+            print(
+                #   f"progress_reward:{progress_reward:.4f}",
+                #   f"lateral_penalty:{lateral_penalty:.4f}",
+                #   f"action_penalty:{action_penalty:.4f}"s, 
+                  f"height_penalty:{height_penalty:.4f}",
+                  f"smoothness_penalty:{smoothness_penalty:.4f},",
+                  f"posture_penalty:{posture_penalty:.4f}",
+                #   f"ang_vel_penalty:{ang_vel_penalty:.4f}",
+                #   f"height_down_penalty:{height_down_penalty:.4f}",
+                #   f"exploration_bonus:{exploration_bonus:.4f}"
+            )
+
+        return total
+
+
+
+    def step(self, action):
+
+        r = self.Player.robot
+        self.previous_action = action
+
+        self.target_joint_positions = (
+                # self.joint_nominal_position +
+                 self.scaling_factor * action
+        )
+        self.target_joint_positions *= self.train_sim_flip
+
+        for idx, target in enumerate(self.target_joint_positions):
+            r.set_motor_target_position(
+                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+            )
+
+        self.previous_action = action
+
+        self.sync()  # run simulation step
+        self.step_counter += 1
+
+        if self.enable_debug_joint_status and self.step_counter % self.debug_every_n_steps == 0:
+            self.debug_joint_status()
+
+        current_pos = np.array(self.Player.world.global_position[:2], dtype=np.float32)
+
+        # Compute reward based on movement from previous step
+        reward = self.compute_reward(self.previous_pos, current_pos, action)
+
+        # Update previous position
+        self.previous_pos = current_pos.copy()
+        self.last_action_for_reward = action.copy()
+
+        # Fall detection and penalty
+        is_fallen = self.Player.world.global_position[2] < 0.3
+
+        # terminal state: the robot is falling or timeout
+        terminated = is_fallen or self.step_counter > 800 or self.route_completed
+        truncated = False
+
+        return self.observe(), reward, terminated, truncated, {}
+
+
+class Train(Train_Base):
+    def __init__(self, script) -> None:
+        super().__init__(script)
+
+    def train(self, args):
+
+        # --------------------------------------- Learning parameters
+        n_envs = 20  # Reduced from 8 to decrease CPU/network pressure during init
+        if n_envs < 1:
+            raise ValueError("GYM_CPU_N_ENVS must be >= 1")
+        n_steps_per_env = 1024  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        minibatch_size = 128  # should be a factor of (n_steps_per_env * n_envs)
+        total_steps = 30000000
+        learning_rate = 3e-4
+        folder_name = f'Walk_R{self.robot_type}'
+        model_path = f'./scripts/gyms/logs/{folder_name}/'
+
+        print(f"Model path: {model_path}")
+        print(f"Using {n_envs} parallel environments")
+
+        # --------------------------------------- Run algorithm
+        def init_env(i_env):
+            def thunk():
+                return WalkEnv(self.ip, self.server_p + i_env)
+
+            return thunk
+
+        server_log_dir = os.path.join(model_path, "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1)  # include 1 extra server for testing
+
+        # Wait for servers to start
+        print(f"Starting {n_envs + 1} rcssservermj servers...")
+        print("Servers started, creating environments...")
+
+        env = SubprocVecEnv([init_env(i) for i in range(n_envs)])
+        eval_env = SubprocVecEnv([init_env(n_envs)])
+
+        try:
+            # Custom policy network architecture
+            policy_kwargs = dict(
+                net_arch=dict(
+                    pi=[512, 256, 128],  # Policy network: 3 layers
+                    vf=[512, 256, 128]  # Value network: 3 layers
+                ),
+                activation_fn=__import__('torch.nn', fromlist=['ELU']).ELU,
+            )
+
+            if "model_file" in args:  # retrain
+                model = PPO.load(args["model_file"], env=env, device="cpu", n_envs=n_envs, n_steps=n_steps_per_env,
+                                 batch_size=minibatch_size, learning_rate=learning_rate)
+            else:  # train new model
+                model = PPO(
+                    "MlpPolicy",
+                    env=env,
+                    verbose=1,
+                    n_steps=n_steps_per_env,
+                    batch_size=minibatch_size,
+                    learning_rate=learning_rate,
+                    device="cpu",
+                    policy_kwargs=policy_kwargs,
+                    ent_coef=0.005,  # Entropy coefficient for exploration
+                    # clip_range=0.13,  # PPO clipping parameter
+                    gae_lambda=0.95,  # GAE lambda
+                    gamma=0.99 , # Discount factor
+                    # target_kl=0.03,
+                    # n_epochs=5
+                )
+
+            model_path = self.learn_model(model, total_steps, model_path, eval_env=eval_env,
+                                          eval_freq=n_steps_per_env * 10, save_freq=n_steps_per_env * 10,
+                                          backup_env_file=__file__)
+        except KeyboardInterrupt:
+            sleep(1)  # wait for child processes
+            print("\nctrl+c pressed, aborting...\n")
+            servers.kill()
+            return
+
+        env.close()
+        eval_env.close()
+        servers.kill()
+
+    def test(self, args):
+
+        # Uses different server and monitor ports
+        server_log_dir = os.path.join(args["folder_dir"], "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        server = Train_Server(self.server_p - 1, self.monitor_p, 1)
+        env = WalkEnv(self.ip, self.server_p - 1)
+        model = PPO.load(args["model_file"], env=env)
+
+        try:
+            self.export_model(args["model_file"], args["model_file"] + ".pkl",
+                              False)  # Export to pkl to create custom behavior
+            self.test_model(model, env, log_path=args["folder_dir"], model_path=args["folder_dir"])
+        except KeyboardInterrupt:
+            print()
+
+        env.close()
+        server.kill()
+
+
+if __name__ == "__main__":
+    from types import SimpleNamespace
+
+    # 创建默认参数
+    script_args = SimpleNamespace(
+        args=SimpleNamespace(
+            i='127.0.0.1',  # Server IP
+            p=3100,  # Server port
+            m=3200,  # Monitor port
+            r=0,  # Robot type
+            t='Gym',  # Team name
+            u=1  # Uniform number
+        )
+    )
+
+    trainer = Train(script_args)
+    trainer.train({})
+    # trainer.test({"model_file": "scripts/gyms/logs/Walk_R0_000/best_model.zip",
+                #   "folder_dir": "scripts/gyms/logs/Walk_R0_000/",})

scripts/gyms/logs/Walk_R0_002/Walk.py

@@ -0,0 +1,625 @@
+import os
+import numpy as np
+import math
+import time
+from time import sleep
+from random import random
+from random import uniform
+
+from stable_baselines3 import PPO
+from stable_baselines3.common.vec_env import SubprocVecEnv
+
+import gymnasium as gym
+from gymnasium import spaces
+
+from scripts.commons.Train_Base import Train_Base
+from scripts.commons.Server import Server as Train_Server
+
+from agent.base_agent import Base_Agent
+from utils.math_ops import MathOps
+
+from scipy.spatial.transform import Rotation as R
+
+'''
+Objective:
+Learn how to run forward using step primitive
+----------
+- class Basic_Run: implements an OpenAI custom gym
+- class Train:  implements algorithms to train a new model or test an existing model
+'''
+
+
+class WalkEnv(gym.Env):
+    def __init__(self, ip, server_p) -> None:
+
+        # Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
+        self.Player = player = Base_Agent(
+            team_name="Gym",
+            number=1,
+            host=ip,
+            port=server_p
+        )
+        self.robot_type = self.Player.robot
+        self.step_counter = 0  # to limit episode size
+        self.force_play_on = True
+
+        self.target_position = np.array([0.0, 0.0])  # target position in the x-y plane
+        self.initial_position = np.array([0.0, 0.0])  # initial position in the x-y plane
+        self.target_direction = 0.0  # target direction in the x-y plane (relative to the robot's orientation)
+        self.isfallen = False
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.debug_every_n_steps = 5
+        self.enable_debug_joint_status = False
+        self.calibrate_nominal_from_neutral = True
+        self.auto_calibrate_train_sim_flip = True
+        self.nominal_calibrated_once = False
+        self.flip_calibrated_once = False
+        self._target_hz = 0.0
+        self._target_dt = 0.0
+        self._last_sync_time = None
+        target_hz_env = 0
+        if target_hz_env:
+            try:
+                self._target_hz = float(target_hz_env)
+            except ValueError:
+                self._target_hz = 0.0
+        if self._target_hz > 0.0:
+            self._target_dt = 1.0 / self._target_hz
+
+        # State space
+        # 原始观测大小: 78
+        obs_size = 78
+        self.obs = np.zeros(obs_size, np.float32)
+        self.observation_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(obs_size,),
+            dtype=np.float32
+        )
+
+        action_dim = len(self.Player.robot.ROBOT_MOTORS)
+        self.no_of_actions = action_dim
+        self.action_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(action_dim,),
+            dtype=np.float32
+        )
+
+        # 中立姿态
+        self.joint_nominal_position = np.array(
+            [
+                0.0,
+                0.0,
+                0.0,
+                1.4,
+                0.0,
+                -0.4,
+                0.0,
+                -1.4,
+                0.0,
+                0.4,
+                0.0,
+                -0.4,
+                0.0,
+                0.0,
+                0.8,
+                -0.4,
+                0.0,
+                0.4,
+                0.0,
+                0.0,
+                -0.8,
+                0.4,
+                0.0,
+            ]
+        )
+        self.joint_nominal_position = np.zeros(self.no_of_actions)
+        self.train_sim_flip = np.array(
+            [
+                1.0,  # 0: Head_yaw (he1)
+                -1.0,  # 1: Head_pitch (he2)
+                1.0,  # 2: Left_Shoulder_Pitch (lae1)
+                -1.0,  # 3: Left_Shoulder_Roll (lae2)
+                -1.0,  # 4: Left_Elbow_Pitch (lae3)
+                1.0,  # 5: Left_Elbow_Yaw (lae4)
+                -1.0,  # 6: Right_Shoulder_Pitch (rae1)
+                -1.0,  # 7: Right_Shoulder_Roll (rae2)
+                1.0,  # 8: Right_Elbow_Pitch (rae3)
+                1.0,  # 9: Right_Elbow_Yaw (rae4)
+                1.0,  # 10: Waist (te1)
+                1.0,  # 11: Left_Hip_Pitch (lle1)
+                -1.0,  # 12: Left_Hip_Roll (lle2)
+                -1.0,  # 13: Left_Hip_Yaw (lle3)
+                1.0,  # 14: Left_Knee_Pitch (lle4)
+                1.0,  # 15: Left_Ankle_Pitch (lle5)
+                -1.0,  # 16: Left_Ankle_Roll (lle6)
+                -1.0,  # 17: Right_Hip_Pitch (rle1)
+                -1.0,  # 18: Right_Hip_Roll (rle2)
+                -1.0,  # 19: Right_Hip_Yaw (rle3)
+                -1.0,  # 20: Right_Knee_Pitch (rle4)
+                -1.0,  # 21: Right_Ankle_Pitch (rle5)
+                -1.0,  # 22: Right_Ankle_Roll (rle6)
+            ]
+        )
+
+        self.scaling_factor = 0.1
+        # self.scaling_factor = 1
+
+        # Small reset perturbations for robustness training.
+        self.enable_reset_perturb = False
+        self.reset_beam_yaw_range_deg = 180  # randomize target direction fully to encourage learning a real walk instead of a fixed gait
+        self.reset_joint_noise_rad = 0.035
+        self.reset_perturb_steps = 5
+        self.reset_recover_steps = 8
+
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Track previous position
+        self.Player.server.connect()
+        # sleep(2.0)  # Longer wait for connection to establish completely
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+        self.start_time = time.time()
+
+    def debug_log(self, message):
+        print(message)
+        try:
+            log_path = os.path.join(os.path.dirname(os.path.dirname(__file__)), "comm_debug.log")
+            with open(log_path, "a", encoding="utf-8") as f:
+                f.write(message + "\n")
+        except OSError:
+            pass
+
+    def observe(self, init=False):
+
+        """获取当前观测值"""
+        robot = self.Player.robot
+        world = self.Player.world
+
+        # Safety check: ensure data is available
+
+        # 计算目标速度
+        raw_target = self.target_position - world.global_position[:2]
+        velocity = MathOps.rotate_2d_vec(
+            raw_target,
+            -robot.global_orientation_euler[2],
+            is_rad=False
+        )
+
+        # 计算相对方向
+        rel_orientation = MathOps.vector_angle(velocity) * 0.3
+        rel_orientation = np.clip(rel_orientation, -0.25, 0.25)
+
+        velocity = np.concatenate([velocity, np.array([rel_orientation])])
+        velocity[0] = np.clip(velocity[0], -0.5, 0.5)
+        velocity[1] = np.clip(velocity[1], -0.25, 0.25)
+
+        # 关节状态
+        radian_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        radian_joint_speeds = np.deg2rad(
+            [robot.motor_speeds[motor] for motor in robot.ROBOT_MOTORS]
+        )
+
+        qpos_qvel_previous_action = np.concatenate([
+            (radian_joint_positions * self.train_sim_flip - self.joint_nominal_position) / 4.6,
+            radian_joint_speeds / 110.0 * self.train_sim_flip,
+            self.previous_action / 10.0,
+        ])
+
+        # 角速度
+        ang_vel = np.clip(np.deg2rad(robot.gyroscope) / 50.0, -1.0, 1.0)
+
+        # 投影的重力方向
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+
+        # 组合观测
+        observation = np.concatenate([
+            qpos_qvel_previous_action,
+            ang_vel,
+            velocity,
+            projected_gravity,
+        ])
+
+        observation = np.clip(observation, -10.0, 10.0)
+        return observation.astype(np.float32)
+
+    def sync(self):
+        ''' Run a single simulation step '''
+        self.Player.server.receive()
+        self.Player.world.update()
+        self.Player.robot.commit_motor_targets_pd()
+        self.Player.server.send()
+        if self._target_dt > 0.0:
+            now = time.time()
+            if self._last_sync_time is None:
+                self._last_sync_time = now
+                return
+            elapsed = now - self._last_sync_time
+            remaining = self._target_dt - elapsed
+            if remaining > 0.0:
+                time.sleep(remaining)
+                now = time.time()
+            self._last_sync_time = now
+
+    def debug_joint_status(self):
+        robot = self.Player.robot
+        actual_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        target_joint_positions = getattr(
+            self,
+            'target_joint_positions',
+            np.zeros(len(robot.ROBOT_MOTORS), dtype=np.float32)
+        )
+        joint_error = actual_joint_positions - target_joint_positions
+        leg_slice = slice(11, None)
+
+        self.debug_log(
+            "[WalkDebug] "
+            f"step={self.step_counter} "
+            f"pos={np.round(self.Player.world.global_position, 3).tolist()} "
+            f"target_xy={np.round(self.target_position, 3).tolist()} "
+            f"target_leg={np.round(target_joint_positions[leg_slice], 3).tolist()} "
+            f"actual_leg={np.round(actual_joint_positions[leg_slice], 3).tolist()} "
+            f"err_norm={float(np.linalg.norm(joint_error)):.4f} "
+            f"fallen={self.Player.world.global_position[2] < 0.3}"
+        )
+        print(f"waist target={target_joint_positions[10]:.3f}, actual={actual_joint_positions[10]:.3f}")
+
+    def reset(self, seed=None, options=None):
+        '''
+        Reset and stabilize the robot
+        Note: for some behaviors it would be better to reduce stabilization or add noise
+        '''
+        r = self.Player.robot
+        super().reset(seed=seed)
+        if seed is not None:
+            np.random.seed(seed)
+
+        length1 = 2  # randomize target distance
+        length2 = np.random.uniform(0.6, 1)  # randomize target distance
+        length3 = np.random.uniform(0.6, 1)  # randomize target distance
+        angle2 = np.random.uniform(-30, 30)  # randomize initial orientation
+        angle3 = np.random.uniform(-30, 30)  # randomize target direction
+
+        self.step_counter = 0
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Initialize for first step
+        self.walk_cycle_step = 0
+
+        # 随机 beam 目标位置和朝向，增加训练多样性
+        beam_x = (random() - 0.5) * 10
+        beam_y = (random() - 0.5) * 10
+        beam_yaw = uniform(-self.reset_beam_yaw_range_deg, self.reset_beam_yaw_range_deg)
+
+        for _ in range(5):
+            self.Player.server.receive()
+            self.Player.world.update()
+            self.Player.robot.commit_motor_targets_pd()
+            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=beam_yaw)
+            self.Player.server.send()
+
+        # 执行 Neutral 技能直到完成，给机器人足够时间在 beam 位置稳定站立
+        finished_count = 0
+        for _ in range(50):
+            finished = self.Player.skills_manager.execute("Neutral")
+            self.sync()
+            if finished:
+                finished_count += 1
+                if finished_count >= 20:  # 假设需要连续20次完成才算成功
+                    break
+
+        if self.enable_reset_perturb and self.reset_joint_noise_rad > 0.0:
+            perturb_action = np.zeros(self.no_of_actions, dtype=np.float32)
+            # Perturb waist + lower body only (10:), keep head/arms stable.
+            perturb_action[10:] = np.random.uniform(
+                -self.reset_joint_noise_rad,
+                self.reset_joint_noise_rad,
+                size=(self.no_of_actions - 10,)
+            )
+
+            for _ in range(self.reset_perturb_steps):
+                target_joint_positions = (self.joint_nominal_position + perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+            for i in range(self.reset_recover_steps):
+                # Linearly fade perturbation to help policy start from near-neutral.
+                alpha = 1.0 - float(i + 1) / float(self.reset_recover_steps)
+                target_joint_positions = (self.joint_nominal_position + alpha * perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+        # memory variables
+        self.sync()  
+        self.initial_position = np.array(self.Player.world.global_position[:2])
+        self.previous_pos = self.initial_position.copy()  # Critical: set to actual position
+        self.act = np.zeros(self.no_of_actions, np.float32)
+        # Build target in the robot's current forward direction instead of fixed global +x.
+        heading_deg = float(r.global_orientation_euler[2])
+        forward_offset = MathOps.rotate_2d_vec(np.array([length1, 0.0]), heading_deg, is_rad=False)
+        point1 = self.initial_position + forward_offset
+        point2 = point1 + MathOps.rotate_2d_vec(np.array([length2, 0]), angle2, is_rad=False)
+        point3 = point2 + MathOps.rotate_2d_vec(np.array([length3, 0]), angle3, is_rad=False)
+        self.point_list = [point1]
+        self.target_position = self.point_list[self.waypoint_index]
+        self.initial_height = self.Player.world.global_position[2]
+
+        return self.observe(True), {}
+
+    def render(self, mode='human', close=False):
+        return
+
+    def compute_reward(self, previous_pos, current_pos, action):
+        height = float(self.Player.world.global_position[2])
+
+        orientation_quat_inv = R.from_quat(self.Player.robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+        tilt_mag = float(np.linalg.norm(projected_gravity[:2]))
+        ang_vel = np.deg2rad(self.Player.robot.gyroscope)
+        ang_vel_mag = float(np.linalg.norm(ang_vel))
+
+        is_fallen = height < 0.3
+        if is_fallen:
+            remain = max(0, 800 - self.step_counter)
+            return -8.0 - 0.01 * remain
+        
+
+
+        # # 目标方向
+        # to_target = self.target_position - current_pos
+        # dist_to_target = float(np.linalg.norm(to_target))
+        # if dist_to_target < 0.5:
+        #     return 15.0
+
+        # forward_dir = to_target / dist_to_target if dist_to_target > 0.1 else np.array([1.0, 0.0])
+        # delta_pos = current_pos - previous_pos
+        # forward_step = float(np.dot(delta_pos, forward_dir))
+        # lateral_step = float(np.linalg.norm(delta_pos - forward_dir * forward_step))
+
+        # 奖励项
+        # progress_reward = 2 * forward_step
+        # lateral_penalty = -0.1 * lateral_step
+        alive_bonus = 0.3
+
+        # action_penalty = -0.01 * float(np.linalg.norm(action))
+        smoothness_penalty = -0.03 * float(np.linalg.norm(action - self.last_action_for_reward))
+
+        posture_penalty = -1.0 * (tilt_mag)
+        # ang_vel_penalty = -0.05 * ang_vel_mag
+
+        target_height = self.initial_height
+        height_error =  height - target_height if abs(height - target_height) > 0.05 else 0.0
+        height_penalty = -2.0 * abs(height_error)  # 惩罚高度偏离，系数可调
+
+        # # 在 compute_reward 开头附近，添加高度变化率计算
+        # if not hasattr(self, 'last_height'):
+        #     self.last_height = height
+        #     self.last_height_time = self.step_counter  # 可选，用于时间间隔
+        # height_rate = height - self.last_height  # 正为上升，负为下降
+        # self.last_height = height
+
+        # 惩罚高度下降（负变化率）
+        # height_down_penalty = -5.0 * max(0, -height_rate)  # 系数可调，-height_rate 为正表示下降幅度
+
+        # # 在 compute_reward 中
+        # if self.step_counter > 50:
+        #     avg_prev_action = np.mean(self.prev_action_history, axis=0)
+        #     novelty = float(np.linalg.norm(action - avg_prev_action))
+        #     exploration_bonus = 0.05 * novelty
+        # else:
+        #     exploration_bonus = 0
+
+        # self.prev_action_history[self.history_idx] = action
+        # self.history_idx = (self.history_idx + 1) % 50
+
+
+        total = (
+            # progress_reward  + 
+                 alive_bonus + 
+                 # lateral_penalty +
+                 # action_penalty +
+                smoothness_penalty +
+                 posture_penalty
+                #  + ang_vel_penalty
+                 + height_penalty
+                #  + exploration_bonus
+                # + height_down_penalty
+                 )
+        if time.time() - self.start_time >= 1200:
+            self.start_time = time.time()
+            print(
+                #   f"progress_reward:{progress_reward:.4f}",
+                #   f"lateral_penalty:{lateral_penalty:.4f}",
+                #   f"action_penalty:{action_penalty:.4f}"s, 
+                  f"height_penalty:{height_penalty:.4f}",
+                  f"smoothness_penalty:{smoothness_penalty:.4f},",
+                  f"posture_penalty:{posture_penalty:.4f}",
+                #   f"ang_vel_penalty:{ang_vel_penalty:.4f}",
+                #   f"height_down_penalty:{height_down_penalty:.4f}",
+                #   f"exploration_bonus:{exploration_bonus:.4f}"
+            )
+
+        return total
+
+
+
+    def step(self, action):
+
+        r = self.Player.robot
+        self.previous_action = action
+
+        self.target_joint_positions = (
+                # self.joint_nominal_position +
+                 self.scaling_factor * action
+        )
+        self.target_joint_positions *= self.train_sim_flip
+
+        for idx, target in enumerate(self.target_joint_positions):
+            r.set_motor_target_position(
+                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+            )
+
+        self.previous_action = action
+
+        self.sync()  # run simulation step
+        self.step_counter += 1
+
+        if self.enable_debug_joint_status and self.step_counter % self.debug_every_n_steps == 0:
+            self.debug_joint_status()
+
+        current_pos = np.array(self.Player.world.global_position[:2], dtype=np.float32)
+
+        # Compute reward based on movement from previous step
+        reward = self.compute_reward(self.previous_pos, current_pos, action)
+
+        # Update previous position
+        self.previous_pos = current_pos.copy()
+        self.last_action_for_reward = action.copy()
+
+        # Fall detection and penalty
+        is_fallen = self.Player.world.global_position[2] < 0.3
+
+        # terminal state: the robot is falling or timeout
+        terminated = is_fallen or self.step_counter > 800 or self.route_completed
+        truncated = False
+
+        return self.observe(), reward, terminated, truncated, {}
+
+
+class Train(Train_Base):
+    def __init__(self, script) -> None:
+        super().__init__(script)
+
+    def train(self, args):
+
+        # --------------------------------------- Learning parameters
+        n_envs = 20  # Reduced from 8 to decrease CPU/network pressure during init
+        if n_envs < 1:
+            raise ValueError("GYM_CPU_N_ENVS must be >= 1")
+        n_steps_per_env = 1024  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        minibatch_size = 128  # should be a factor of (n_steps_per_env * n_envs)
+        total_steps = 30000000
+        learning_rate = 3e-4
+        folder_name = f'Walk_R{self.robot_type}'
+        model_path = f'./scripts/gyms/logs/{folder_name}/'
+
+        print(f"Model path: {model_path}")
+        print(f"Using {n_envs} parallel environments")
+
+        # --------------------------------------- Run algorithm
+        def init_env(i_env):
+            def thunk():
+                return WalkEnv(self.ip, self.server_p + i_env)
+
+            return thunk
+
+        server_log_dir = os.path.join(model_path, "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1)  # include 1 extra server for testing
+
+        # Wait for servers to start
+        print(f"Starting {n_envs + 1} rcssservermj servers...")
+        print("Servers started, creating environments...")
+
+        env = SubprocVecEnv([init_env(i) for i in range(n_envs)])
+        eval_env = SubprocVecEnv([init_env(n_envs)])
+
+        try:
+            # Custom policy network architecture
+            policy_kwargs = dict(
+                net_arch=dict(
+                    pi=[512, 256, 128],  # Policy network: 3 layers
+                    vf=[512, 256, 128]  # Value network: 3 layers
+                ),
+                activation_fn=__import__('torch.nn', fromlist=['ELU']).ELU,
+            )
+
+            if "model_file" in args:  # retrain
+                model = PPO.load(args["model_file"], env=env, device="cpu", n_envs=n_envs, n_steps=n_steps_per_env,
+                                 batch_size=minibatch_size, learning_rate=learning_rate)
+            else:  # train new model
+                model = PPO(
+                    "MlpPolicy",
+                    env=env,
+                    verbose=1,
+                    n_steps=n_steps_per_env,
+                    batch_size=minibatch_size,
+                    learning_rate=learning_rate,
+                    device="cpu",
+                    policy_kwargs=policy_kwargs,
+                    ent_coef=0.005,  # Entropy coefficient for exploration
+                    # clip_range=0.13,  # PPO clipping parameter
+                    gae_lambda=0.95,  # GAE lambda
+                    gamma=0.99 , # Discount factor
+                    # target_kl=0.03,
+                    # n_epochs=5
+                )
+
+            model_path = self.learn_model(model, total_steps, model_path, eval_env=eval_env,
+                                          eval_freq=n_steps_per_env * 10, save_freq=n_steps_per_env * 10,
+                                          backup_env_file=__file__)
+        except KeyboardInterrupt:
+            sleep(1)  # wait for child processes
+            print("\nctrl+c pressed, aborting...\n")
+            servers.kill()
+            return
+
+        env.close()
+        eval_env.close()
+        servers.kill()
+
+    def test(self, args):
+
+        # Uses different server and monitor ports
+        server_log_dir = os.path.join(args["folder_dir"], "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        server = Train_Server(self.server_p - 1, self.monitor_p, 1)
+        env = WalkEnv(self.ip, self.server_p - 1)
+        model = PPO.load(args["model_file"], env=env)
+
+        try:
+            self.export_model(args["model_file"], args["model_file"] + ".pkl",
+                              False)  # Export to pkl to create custom behavior
+            self.test_model(model, env, log_path=args["folder_dir"], model_path=args["folder_dir"])
+        except KeyboardInterrupt:
+            print()
+
+        env.close()
+        server.kill()
+
+
+if __name__ == "__main__":
+    from types import SimpleNamespace
+
+    # 创建默认参数
+    script_args = SimpleNamespace(
+        args=SimpleNamespace(
+            i='127.0.0.1',  # Server IP
+            p=3100,  # Server port
+            m=3200,  # Monitor port
+            r=0,  # Robot type
+            t='Gym',  # Team name
+            u=1  # Uniform number
+        )
+    )
+
+    trainer = Train(script_args)
+    trainer.train({})
+    # trainer.test({"model_file": "scripts/gyms/logs/Walk_R0_000/best_model.zip",
+                #   "folder_dir": "scripts/gyms/logs/Walk_R0_000/",})

scripts/gyms/logs/Walk_R0_003/Walk.py

@@ -0,0 +1,625 @@
+import os
+import numpy as np
+import math
+import time
+from time import sleep
+from random import random
+from random import uniform
+
+from stable_baselines3 import PPO
+from stable_baselines3.common.vec_env import SubprocVecEnv
+
+import gymnasium as gym
+from gymnasium import spaces
+
+from scripts.commons.Train_Base import Train_Base
+from scripts.commons.Server import Server as Train_Server
+
+from agent.base_agent import Base_Agent
+from utils.math_ops import MathOps
+
+from scipy.spatial.transform import Rotation as R
+
+'''
+Objective:
+Learn how to run forward using step primitive
+----------
+- class Basic_Run: implements an OpenAI custom gym
+- class Train:  implements algorithms to train a new model or test an existing model
+'''
+
+
+class WalkEnv(gym.Env):
+    def __init__(self, ip, server_p) -> None:
+
+        # Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
+        self.Player = player = Base_Agent(
+            team_name="Gym",
+            number=1,
+            host=ip,
+            port=server_p
+        )
+        self.robot_type = self.Player.robot
+        self.step_counter = 0  # to limit episode size
+        self.force_play_on = True
+
+        self.target_position = np.array([0.0, 0.0])  # target position in the x-y plane
+        self.initial_position = np.array([0.0, 0.0])  # initial position in the x-y plane
+        self.target_direction = 0.0  # target direction in the x-y plane (relative to the robot's orientation)
+        self.isfallen = False
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.debug_every_n_steps = 5
+        self.enable_debug_joint_status = False
+        self.calibrate_nominal_from_neutral = True
+        self.auto_calibrate_train_sim_flip = True
+        self.nominal_calibrated_once = False
+        self.flip_calibrated_once = False
+        self._target_hz = 0.0
+        self._target_dt = 0.0
+        self._last_sync_time = None
+        target_hz_env = 0
+        if target_hz_env:
+            try:
+                self._target_hz = float(target_hz_env)
+            except ValueError:
+                self._target_hz = 0.0
+        if self._target_hz > 0.0:
+            self._target_dt = 1.0 / self._target_hz
+
+        # State space
+        # 原始观测大小: 78
+        obs_size = 78
+        self.obs = np.zeros(obs_size, np.float32)
+        self.observation_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(obs_size,),
+            dtype=np.float32
+        )
+
+        action_dim = len(self.Player.robot.ROBOT_MOTORS)
+        self.no_of_actions = action_dim
+        self.action_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(action_dim,),
+            dtype=np.float32
+        )
+
+        # 中立姿态
+        self.joint_nominal_position = np.array(
+            [
+                0.0,
+                0.0,
+                0.0,
+                1.4,
+                0.0,
+                -0.4,
+                0.0,
+                -1.4,
+                0.0,
+                0.4,
+                0.0,
+                -0.4,
+                0.0,
+                0.0,
+                0.8,
+                -0.4,
+                0.0,
+                0.4,
+                0.0,
+                0.0,
+                -0.8,
+                0.4,
+                0.0,
+            ]
+        )
+        self.joint_nominal_position = np.zeros(self.no_of_actions)
+        self.train_sim_flip = np.array(
+            [
+                1.0,  # 0: Head_yaw (he1)
+                -1.0,  # 1: Head_pitch (he2)
+                1.0,  # 2: Left_Shoulder_Pitch (lae1)
+                -1.0,  # 3: Left_Shoulder_Roll (lae2)
+                -1.0,  # 4: Left_Elbow_Pitch (lae3)
+                1.0,  # 5: Left_Elbow_Yaw (lae4)
+                -1.0,  # 6: Right_Shoulder_Pitch (rae1)
+                -1.0,  # 7: Right_Shoulder_Roll (rae2)
+                1.0,  # 8: Right_Elbow_Pitch (rae3)
+                1.0,  # 9: Right_Elbow_Yaw (rae4)
+                1.0,  # 10: Waist (te1)
+                1.0,  # 11: Left_Hip_Pitch (lle1)
+                -1.0,  # 12: Left_Hip_Roll (lle2)
+                -1.0,  # 13: Left_Hip_Yaw (lle3)
+                1.0,  # 14: Left_Knee_Pitch (lle4)
+                1.0,  # 15: Left_Ankle_Pitch (lle5)
+                -1.0,  # 16: Left_Ankle_Roll (lle6)
+                -1.0,  # 17: Right_Hip_Pitch (rle1)
+                -1.0,  # 18: Right_Hip_Roll (rle2)
+                -1.0,  # 19: Right_Hip_Yaw (rle3)
+                -1.0,  # 20: Right_Knee_Pitch (rle4)
+                -1.0,  # 21: Right_Ankle_Pitch (rle5)
+                -1.0,  # 22: Right_Ankle_Roll (rle6)
+            ]
+        )
+
+        self.scaling_factor = 0.3
+        # self.scaling_factor = 1
+
+        # Small reset perturbations for robustness training.
+        self.enable_reset_perturb = True
+        self.reset_beam_yaw_range_deg = 180  # randomize target direction fully to encourage learning a real walk instead of a fixed gait
+        self.reset_joint_noise_rad = 0.035
+        self.reset_perturb_steps = 5
+        self.reset_recover_steps = 8
+
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Track previous position
+        self.Player.server.connect()
+        # sleep(2.0)  # Longer wait for connection to establish completely
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+        self.start_time = time.time()
+
+    def debug_log(self, message):
+        print(message)
+        try:
+            log_path = os.path.join(os.path.dirname(os.path.dirname(__file__)), "comm_debug.log")
+            with open(log_path, "a", encoding="utf-8") as f:
+                f.write(message + "\n")
+        except OSError:
+            pass
+
+    def observe(self, init=False):
+
+        """获取当前观测值"""
+        robot = self.Player.robot
+        world = self.Player.world
+
+        # Safety check: ensure data is available
+
+        # 计算目标速度
+        raw_target = self.target_position - world.global_position[:2]
+        velocity = MathOps.rotate_2d_vec(
+            raw_target,
+            -robot.global_orientation_euler[2],
+            is_rad=False
+        )
+
+        # 计算相对方向
+        rel_orientation = MathOps.vector_angle(velocity) * 0.3
+        rel_orientation = np.clip(rel_orientation, -0.25, 0.25)
+
+        velocity = np.concatenate([velocity, np.array([rel_orientation])])
+        velocity[0] = np.clip(velocity[0], -0.5, 0.5)
+        velocity[1] = np.clip(velocity[1], -0.25, 0.25)
+
+        # 关节状态
+        radian_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        radian_joint_speeds = np.deg2rad(
+            [robot.motor_speeds[motor] for motor in robot.ROBOT_MOTORS]
+        )
+
+        qpos_qvel_previous_action = np.concatenate([
+            (radian_joint_positions * self.train_sim_flip - self.joint_nominal_position) / 4.6,
+            radian_joint_speeds / 110.0 * self.train_sim_flip,
+            self.previous_action / 10.0,
+        ])
+
+        # 角速度
+        ang_vel = np.clip(np.deg2rad(robot.gyroscope) / 50.0, -1.0, 1.0)
+
+        # 投影的重力方向
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+
+        # 组合观测
+        observation = np.concatenate([
+            qpos_qvel_previous_action,
+            ang_vel,
+            velocity,
+            projected_gravity,
+        ])
+
+        observation = np.clip(observation, -10.0, 10.0)
+        return observation.astype(np.float32)
+
+    def sync(self):
+        ''' Run a single simulation step '''
+        self.Player.server.receive()
+        self.Player.world.update()
+        self.Player.robot.commit_motor_targets_pd()
+        self.Player.server.send()
+        if self._target_dt > 0.0:
+            now = time.time()
+            if self._last_sync_time is None:
+                self._last_sync_time = now
+                return
+            elapsed = now - self._last_sync_time
+            remaining = self._target_dt - elapsed
+            if remaining > 0.0:
+                time.sleep(remaining)
+                now = time.time()
+            self._last_sync_time = now
+
+    def debug_joint_status(self):
+        robot = self.Player.robot
+        actual_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        target_joint_positions = getattr(
+            self,
+            'target_joint_positions',
+            np.zeros(len(robot.ROBOT_MOTORS), dtype=np.float32)
+        )
+        joint_error = actual_joint_positions - target_joint_positions
+        leg_slice = slice(11, None)
+
+        self.debug_log(
+            "[WalkDebug] "
+            f"step={self.step_counter} "
+            f"pos={np.round(self.Player.world.global_position, 3).tolist()} "
+            f"target_xy={np.round(self.target_position, 3).tolist()} "
+            f"target_leg={np.round(target_joint_positions[leg_slice], 3).tolist()} "
+            f"actual_leg={np.round(actual_joint_positions[leg_slice], 3).tolist()} "
+            f"err_norm={float(np.linalg.norm(joint_error)):.4f} "
+            f"fallen={self.Player.world.global_position[2] < 0.3}"
+        )
+        print(f"waist target={target_joint_positions[10]:.3f}, actual={actual_joint_positions[10]:.3f}")
+
+    def reset(self, seed=None, options=None):
+        '''
+        Reset and stabilize the robot
+        Note: for some behaviors it would be better to reduce stabilization or add noise
+        '''
+        r = self.Player.robot
+        super().reset(seed=seed)
+        if seed is not None:
+            np.random.seed(seed)
+
+        length1 = 2  # randomize target distance
+        length2 = np.random.uniform(0.6, 1)  # randomize target distance
+        length3 = np.random.uniform(0.6, 1)  # randomize target distance
+        angle2 = np.random.uniform(-30, 30)  # randomize initial orientation
+        angle3 = np.random.uniform(-30, 30)  # randomize target direction
+
+        self.step_counter = 0
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Initialize for first step
+        self.walk_cycle_step = 0
+
+        # 随机 beam 目标位置和朝向，增加训练多样性
+        beam_x = (random() - 0.5) * 10
+        beam_y = (random() - 0.5) * 10
+        beam_yaw = uniform(-self.reset_beam_yaw_range_deg, self.reset_beam_yaw_range_deg)
+
+        for _ in range(5):
+            self.Player.server.receive()
+            self.Player.world.update()
+            self.Player.robot.commit_motor_targets_pd()
+            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=beam_yaw)
+            self.Player.server.send()
+
+        # 执行 Neutral 技能直到完成，给机器人足够时间在 beam 位置稳定站立
+        finished_count = 0
+        for _ in range(50):
+            finished = self.Player.skills_manager.execute("Neutral")
+            self.sync()
+            if finished:
+                finished_count += 1
+                if finished_count >= 20:  # 假设需要连续20次完成才算成功
+                    break
+
+        if self.enable_reset_perturb and self.reset_joint_noise_rad > 0.0:
+            perturb_action = np.zeros(self.no_of_actions, dtype=np.float32)
+            # Perturb waist + lower body only (10:), keep head/arms stable.
+            perturb_action[10:] = np.random.uniform(
+                -self.reset_joint_noise_rad,
+                self.reset_joint_noise_rad,
+                size=(self.no_of_actions - 10,)
+            )
+
+            for _ in range(self.reset_perturb_steps):
+                target_joint_positions = (self.joint_nominal_position + perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+            for i in range(self.reset_recover_steps):
+                # Linearly fade perturbation to help policy start from near-neutral.
+                alpha = 1.0 - float(i + 1) / float(self.reset_recover_steps)
+                target_joint_positions = (self.joint_nominal_position + alpha * perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+        # memory variables
+        self.sync()  
+        self.initial_position = np.array(self.Player.world.global_position[:2])
+        self.previous_pos = self.initial_position.copy()  # Critical: set to actual position
+        self.act = np.zeros(self.no_of_actions, np.float32)
+        # Build target in the robot's current forward direction instead of fixed global +x.
+        heading_deg = float(r.global_orientation_euler[2])
+        forward_offset = MathOps.rotate_2d_vec(np.array([length1, 0.0]), heading_deg, is_rad=False)
+        point1 = self.initial_position + forward_offset
+        point2 = point1 + MathOps.rotate_2d_vec(np.array([length2, 0]), angle2, is_rad=False)
+        point3 = point2 + MathOps.rotate_2d_vec(np.array([length3, 0]), angle3, is_rad=False)
+        self.point_list = [point1]
+        self.target_position = self.point_list[self.waypoint_index]
+        self.initial_height = self.Player.world.global_position[2]
+
+        return self.observe(True), {}
+
+    def render(self, mode='human', close=False):
+        return
+
+    def compute_reward(self, previous_pos, current_pos, action):
+        height = float(self.Player.world.global_position[2])
+
+        orientation_quat_inv = R.from_quat(self.Player.robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+        tilt_mag = float(np.linalg.norm(projected_gravity[:2]))
+        ang_vel = np.deg2rad(self.Player.robot.gyroscope)
+        ang_vel_mag = float(np.linalg.norm(ang_vel))
+
+        is_fallen = height < 0.3
+        if is_fallen:
+            remain = max(0, 800 - self.step_counter)
+            return -8.0 - 0.01 * remain
+        
+
+
+        # # 目标方向
+        # to_target = self.target_position - current_pos
+        # dist_to_target = float(np.linalg.norm(to_target))
+        # if dist_to_target < 0.5:
+        #     return 15.0
+
+        # forward_dir = to_target / dist_to_target if dist_to_target > 0.1 else np.array([1.0, 0.0])
+        # delta_pos = current_pos - previous_pos
+        # forward_step = float(np.dot(delta_pos, forward_dir))
+        # lateral_step = float(np.linalg.norm(delta_pos - forward_dir * forward_step))
+
+        # 奖励项
+        # progress_reward = 2 * forward_step
+        # lateral_penalty = -0.1 * lateral_step
+        alive_bonus = 0.3
+
+        # action_penalty = -0.01 * float(np.linalg.norm(action))
+        smoothness_penalty = -0.03 * float(np.linalg.norm(action - self.last_action_for_reward))
+
+        posture_penalty = -1.0 * (tilt_mag)
+        ang_vel_penalty = -0.05 * ang_vel_mag
+
+        target_height = self.initial_height
+        height_error =  height - target_height
+        height_penalty = -2.0 * abs(height_error)  # 惩罚高度偏离，系数可调
+
+        # # 在 compute_reward 开头附近，添加高度变化率计算
+        # if not hasattr(self, 'last_height'):
+        #     self.last_height = height
+        #     self.last_height_time = self.step_counter  # 可选，用于时间间隔
+        # height_rate = height - self.last_height  # 正为上升，负为下降
+        # self.last_height = height
+
+        # 惩罚高度下降（负变化率）
+        # height_down_penalty = -5.0 * max(0, -height_rate)  # 系数可调，-height_rate 为正表示下降幅度
+
+        # # 在 compute_reward 中
+        # if self.step_counter > 50:
+        #     avg_prev_action = np.mean(self.prev_action_history, axis=0)
+        #     novelty = float(np.linalg.norm(action - avg_prev_action))
+        #     exploration_bonus = 0.05 * novelty
+        # else:
+        #     exploration_bonus = 0
+
+        # self.prev_action_history[self.history_idx] = action
+        # self.history_idx = (self.history_idx + 1) % 50
+
+
+        total = (
+            # progress_reward  + 
+                 alive_bonus + 
+                 # lateral_penalty +
+                 # action_penalty +
+                smoothness_penalty +
+                 posture_penalty
+                 + ang_vel_penalty
+                 + height_penalty
+                #  + exploration_bonus
+                # + height_down_penalty
+                 )
+        if time.time() - self.start_time >= 1200:
+            self.start_time = time.time()
+            print(
+                #   f"progress_reward:{progress_reward:.4f}",
+                #   f"lateral_penalty:{lateral_penalty:.4f}",
+                #   f"action_penalty:{action_penalty:.4f}"s, 
+                  f"height_penalty:{height_penalty:.4f}",
+                  f"smoothness_penalty:{smoothness_penalty:.4f},",
+                  f"posture_penalty:{posture_penalty:.4f}",
+                #   f"ang_vel_penalty:{ang_vel_penalty:.4f}",
+                #   f"height_down_penalty:{height_down_penalty:.4f}",
+                #   f"exploration_bonus:{exploration_bonus:.4f}"
+            )
+
+        return total
+
+
+
+    def step(self, action):
+
+        r = self.Player.robot
+        self.previous_action = action
+
+        self.target_joint_positions = (
+                # self.joint_nominal_position +
+                 self.scaling_factor * action
+        )
+        self.target_joint_positions *= self.train_sim_flip
+
+        for idx, target in enumerate(self.target_joint_positions):
+            r.set_motor_target_position(
+                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+            )
+
+        self.previous_action = action
+
+        self.sync()  # run simulation step
+        self.step_counter += 1
+
+        if self.enable_debug_joint_status and self.step_counter % self.debug_every_n_steps == 0:
+            self.debug_joint_status()
+
+        current_pos = np.array(self.Player.world.global_position[:2], dtype=np.float32)
+
+        # Compute reward based on movement from previous step
+        reward = self.compute_reward(self.previous_pos, current_pos, action)
+
+        # Update previous position
+        self.previous_pos = current_pos.copy()
+        self.last_action_for_reward = action.copy()
+
+        # Fall detection and penalty
+        is_fallen = self.Player.world.global_position[2] < 0.3
+
+        # terminal state: the robot is falling or timeout
+        terminated = is_fallen or self.step_counter > 800 or self.route_completed
+        truncated = False
+
+        return self.observe(), reward, terminated, truncated, {}
+
+
+class Train(Train_Base):
+    def __init__(self, script) -> None:
+        super().__init__(script)
+
+    def train(self, args):
+
+        # --------------------------------------- Learning parameters
+        n_envs = 20  # Reduced from 8 to decrease CPU/network pressure during init
+        if n_envs < 1:
+            raise ValueError("GYM_CPU_N_ENVS must be >= 1")
+        n_steps_per_env = 1024  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        minibatch_size = 128  # should be a factor of (n_steps_per_env * n_envs)
+        total_steps = 30000000
+        learning_rate = 1e-4
+        folder_name = f'Walk_R{self.robot_type}'
+        model_path = f'./scripts/gyms/logs/{folder_name}/'
+
+        print(f"Model path: {model_path}")
+        print(f"Using {n_envs} parallel environments")
+
+        # --------------------------------------- Run algorithm
+        def init_env(i_env):
+            def thunk():
+                return WalkEnv(self.ip, self.server_p + i_env)
+
+            return thunk
+
+        server_log_dir = os.path.join(model_path, "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1)  # include 1 extra server for testing
+
+        # Wait for servers to start
+        print(f"Starting {n_envs + 1} rcssservermj servers...")
+        print("Servers started, creating environments...")
+
+        env = SubprocVecEnv([init_env(i) for i in range(n_envs)])
+        eval_env = SubprocVecEnv([init_env(n_envs)])
+
+        try:
+            # Custom policy network architecture
+            policy_kwargs = dict(
+                net_arch=dict(
+                    pi=[512, 256, 128],  # Policy network: 3 layers
+                    vf=[512, 256, 128]  # Value network: 3 layers
+                ),
+                activation_fn=__import__('torch.nn', fromlist=['ELU']).ELU,
+            )
+
+            if "model_file" in args:  # retrain
+                model = PPO.load(args["model_file"], env=env, device="cpu", n_envs=n_envs, n_steps=n_steps_per_env,
+                                 batch_size=minibatch_size, learning_rate=learning_rate)
+            else:  # train new model
+                model = PPO(
+                    "MlpPolicy",
+                    env=env,
+                    verbose=1,
+                    n_steps=n_steps_per_env,
+                    batch_size=minibatch_size,
+                    learning_rate=learning_rate,
+                    device="cpu",
+                    policy_kwargs=policy_kwargs,
+                    ent_coef=0.001,  # Entropy coefficient for exploration
+                    # clip_range=0.13,  # PPO clipping parameter
+                    gae_lambda=0.95,  # GAE lambda
+                    gamma=0.99 , # Discount factor
+                    target_kl=0.03,
+                    # n_epochs=5
+                )
+
+            model_path = self.learn_model(model, total_steps, model_path, eval_env=eval_env,
+                                          eval_freq=n_steps_per_env * 10, save_freq=n_steps_per_env * 10,
+                                          backup_env_file=__file__)
+        except KeyboardInterrupt:
+            sleep(1)  # wait for child processes
+            print("\nctrl+c pressed, aborting...\n")
+            servers.kill()
+            return
+
+        env.close()
+        eval_env.close()
+        servers.kill()
+
+    def test(self, args):
+
+        # Uses different server and monitor ports
+        server_log_dir = os.path.join(args["folder_dir"], "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        server = Train_Server(self.server_p - 1, self.monitor_p, 1)
+        env = WalkEnv(self.ip, self.server_p - 1)
+        model = PPO.load(args["model_file"], env=env)
+
+        try:
+            self.export_model(args["model_file"], args["model_file"] + ".pkl",
+                              False)  # Export to pkl to create custom behavior
+            self.test_model(model, env, log_path=args["folder_dir"], model_path=args["folder_dir"])
+        except KeyboardInterrupt:
+            print()
+
+        env.close()
+        server.kill()
+
+
+if __name__ == "__main__":
+    from types import SimpleNamespace
+
+    # 创建默认参数
+    script_args = SimpleNamespace(
+        args=SimpleNamespace(
+            i='127.0.0.1',  # Server IP
+            p=3100,  # Server port
+            m=3200,  # Monitor port
+            r=0,  # Robot type
+            t='Gym',  # Team name
+            u=1  # Uniform number
+        )
+    )
+
+    trainer = Train(script_args)
+    trainer.train({})
+    # trainer.test({"model_file": "scripts/gyms/logs/Walk_R0_000/best_model.zip",
+                #   "folder_dir": "scripts/gyms/logs/Walk_R0_000/",})

scripts/gyms/logs/Walk_R0_004/Walk.py

@@ -0,0 +1,626 @@
+import os
+import numpy as np
+import math
+import time
+from time import sleep
+from random import random
+from random import uniform
+
+from stable_baselines3 import PPO
+from stable_baselines3.common.vec_env import SubprocVecEnv
+
+import gymnasium as gym
+from gymnasium import spaces
+
+from scripts.commons.Train_Base import Train_Base
+from scripts.commons.Server import Server as Train_Server
+
+from agent.base_agent import Base_Agent
+from utils.math_ops import MathOps
+
+from scipy.spatial.transform import Rotation as R
+
+'''
+Objective:
+Learn how to run forward using step primitive
+----------
+- class Basic_Run: implements an OpenAI custom gym
+- class Train:  implements algorithms to train a new model or test an existing model
+'''
+
+
+class WalkEnv(gym.Env):
+    def __init__(self, ip, server_p) -> None:
+
+        # Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
+        self.Player = player = Base_Agent(
+            team_name="Gym",
+            number=1,
+            host=ip,
+            port=server_p
+        )
+        self.robot_type = self.Player.robot
+        self.step_counter = 0  # to limit episode size
+        self.force_play_on = True
+
+        self.target_position = np.array([0.0, 0.0])  # target position in the x-y plane
+        self.initial_position = np.array([0.0, 0.0])  # initial position in the x-y plane
+        self.target_direction = 0.0  # target direction in the x-y plane (relative to the robot's orientation)
+        self.isfallen = False
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.debug_every_n_steps = 5
+        self.enable_debug_joint_status = False
+        self.calibrate_nominal_from_neutral = True
+        self.auto_calibrate_train_sim_flip = True
+        self.nominal_calibrated_once = False
+        self.flip_calibrated_once = False
+        self._target_hz = 0.0
+        self._target_dt = 0.0
+        self._last_sync_time = None
+        target_hz_env = 0
+        if target_hz_env:
+            try:
+                self._target_hz = float(target_hz_env)
+            except ValueError:
+                self._target_hz = 0.0
+        if self._target_hz > 0.0:
+            self._target_dt = 1.0 / self._target_hz
+
+        # State space
+        # 原始观测大小: 78
+        obs_size = 78
+        self.obs = np.zeros(obs_size, np.float32)
+        self.observation_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(obs_size,),
+            dtype=np.float32
+        )
+
+        action_dim = len(self.Player.robot.ROBOT_MOTORS)
+        self.no_of_actions = action_dim
+        self.action_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(action_dim,),
+            dtype=np.float32
+        )
+
+        # 中立姿态
+        self.joint_nominal_position = np.array(
+            [
+                0.0,
+                0.0,
+                0.0,
+                1.4,
+                0.0,
+                -0.4,
+                0.0,
+                -1.4,
+                0.0,
+                0.4,
+                0.0,
+                -0.4,
+                0.0,
+                0.0,
+                0.8,
+                -0.4,
+                0.0,
+                0.4,
+                0.0,
+                0.0,
+                -0.8,
+                0.4,
+                0.0,
+            ]
+        )
+        self.joint_nominal_position = np.zeros(self.no_of_actions)
+        self.train_sim_flip = np.array(
+            [
+                1.0,  # 0: Head_yaw (he1)
+                -1.0,  # 1: Head_pitch (he2)
+                1.0,  # 2: Left_Shoulder_Pitch (lae1)
+                -1.0,  # 3: Left_Shoulder_Roll (lae2)
+                -1.0,  # 4: Left_Elbow_Pitch (lae3)
+                1.0,  # 5: Left_Elbow_Yaw (lae4)
+                -1.0,  # 6: Right_Shoulder_Pitch (rae1)
+                -1.0,  # 7: Right_Shoulder_Roll (rae2)
+                1.0,  # 8: Right_Elbow_Pitch (rae3)
+                1.0,  # 9: Right_Elbow_Yaw (rae4)
+                1.0,  # 10: Waist (te1)
+                1.0,  # 11: Left_Hip_Pitch (lle1)
+                -1.0,  # 12: Left_Hip_Roll (lle2)
+                -1.0,  # 13: Left_Hip_Yaw (lle3)
+                1.0,  # 14: Left_Knee_Pitch (lle4)
+                1.0,  # 15: Left_Ankle_Pitch (lle5)
+                -1.0,  # 16: Left_Ankle_Roll (lle6)
+                -1.0,  # 17: Right_Hip_Pitch (rle1)
+                -1.0,  # 18: Right_Hip_Roll (rle2)
+                -1.0,  # 19: Right_Hip_Yaw (rle3)
+                -1.0,  # 20: Right_Knee_Pitch (rle4)
+                -1.0,  # 21: Right_Ankle_Pitch (rle5)
+                -1.0,  # 22: Right_Ankle_Roll (rle6)
+            ]
+        )
+
+        self.scaling_factor = 0.3
+        # self.scaling_factor = 1
+
+        # Small reset perturbations for robustness training.
+        self.enable_reset_perturb = True
+        self.reset_beam_yaw_range_deg = 180  # randomize target direction fully to encourage learning a real walk instead of a fixed gait
+        self.reset_joint_noise_rad = 0.015
+        self.reset_perturb_steps = 3
+        self.reset_recover_steps = 8
+
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Track previous position
+        self.Player.server.connect()
+        # sleep(2.0)  # Longer wait for connection to establish completely
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+        self.start_time = time.time()
+
+    def debug_log(self, message):
+        print(message)
+        try:
+            log_path = os.path.join(os.path.dirname(os.path.dirname(__file__)), "comm_debug.log")
+            with open(log_path, "a", encoding="utf-8") as f:
+                f.write(message + "\n")
+        except OSError:
+            pass
+
+    def observe(self, init=False):
+
+        """获取当前观测值"""
+        robot = self.Player.robot
+        world = self.Player.world
+
+        # Safety check: ensure data is available
+
+        # 计算目标速度
+        raw_target = self.target_position - world.global_position[:2]
+        velocity = MathOps.rotate_2d_vec(
+            raw_target,
+            -robot.global_orientation_euler[2],
+            is_rad=False
+        )
+
+        # 计算相对方向
+        rel_orientation = MathOps.vector_angle(velocity) * 0.3
+        rel_orientation = np.clip(rel_orientation, -0.25, 0.25)
+
+        velocity = np.concatenate([velocity, np.array([rel_orientation])])
+        velocity[0] = np.clip(velocity[0], -0.5, 0.5)
+        velocity[1] = np.clip(velocity[1], -0.25, 0.25)
+
+        # 关节状态
+        radian_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        radian_joint_speeds = np.deg2rad(
+            [robot.motor_speeds[motor] for motor in robot.ROBOT_MOTORS]
+        )
+
+        qpos_qvel_previous_action = np.concatenate([
+            (radian_joint_positions * self.train_sim_flip - self.joint_nominal_position) / 4.6,
+            radian_joint_speeds / 110.0 * self.train_sim_flip,
+            self.previous_action / 10.0,
+        ])
+
+        # 角速度
+        ang_vel = np.clip(np.deg2rad(robot.gyroscope) / 50.0, -1.0, 1.0)
+
+        # 投影的重力方向
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+
+        # 组合观测
+        observation = np.concatenate([
+            qpos_qvel_previous_action,
+            ang_vel,
+            velocity,
+            projected_gravity,
+        ])
+
+        observation = np.clip(observation, -10.0, 10.0)
+        return observation.astype(np.float32)
+
+    def sync(self):
+        ''' Run a single simulation step '''
+        self.Player.server.receive()
+        self.Player.world.update()
+        self.Player.robot.commit_motor_targets_pd()
+        self.Player.server.send()
+        if self._target_dt > 0.0:
+            now = time.time()
+            if self._last_sync_time is None:
+                self._last_sync_time = now
+                return
+            elapsed = now - self._last_sync_time
+            remaining = self._target_dt - elapsed
+            if remaining > 0.0:
+                time.sleep(remaining)
+                now = time.time()
+            self._last_sync_time = now
+
+    def debug_joint_status(self):
+        robot = self.Player.robot
+        actual_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        target_joint_positions = getattr(
+            self,
+            'target_joint_positions',
+            np.zeros(len(robot.ROBOT_MOTORS), dtype=np.float32)
+        )
+        joint_error = actual_joint_positions - target_joint_positions
+        leg_slice = slice(11, None)
+
+        self.debug_log(
+            "[WalkDebug] "
+            f"step={self.step_counter} "
+            f"pos={np.round(self.Player.world.global_position, 3).tolist()} "
+            f"target_xy={np.round(self.target_position, 3).tolist()} "
+            f"target_leg={np.round(target_joint_positions[leg_slice], 3).tolist()} "
+            f"actual_leg={np.round(actual_joint_positions[leg_slice], 3).tolist()} "
+            f"err_norm={float(np.linalg.norm(joint_error)):.4f} "
+            f"fallen={self.Player.world.global_position[2] < 0.3}"
+        )
+        print(f"waist target={target_joint_positions[10]:.3f}, actual={actual_joint_positions[10]:.3f}")
+
+    def reset(self, seed=None, options=None):
+        '''
+        Reset and stabilize the robot
+        Note: for some behaviors it would be better to reduce stabilization or add noise
+        '''
+        r = self.Player.robot
+        super().reset(seed=seed)
+        if seed is not None:
+            np.random.seed(seed)
+
+        length1 = 2  # randomize target distance
+        length2 = np.random.uniform(0.6, 1)  # randomize target distance
+        length3 = np.random.uniform(0.6, 1)  # randomize target distance
+        angle2 = np.random.uniform(-30, 30)  # randomize initial orientation
+        angle3 = np.random.uniform(-30, 30)  # randomize target direction
+
+        self.step_counter = 0
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Initialize for first step
+        self.walk_cycle_step = 0
+
+        # 随机 beam 目标位置和朝向，增加训练多样性
+        beam_x = (random() - 0.5) * 10
+        beam_y = (random() - 0.5) * 10
+        beam_yaw = uniform(-self.reset_beam_yaw_range_deg, self.reset_beam_yaw_range_deg)
+
+        for _ in range(5):
+            self.Player.server.receive()
+            self.Player.world.update()
+            self.Player.robot.commit_motor_targets_pd()
+            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=beam_yaw)
+            self.Player.server.send()
+
+        # 执行 Neutral 技能直到完成，给机器人足够时间在 beam 位置稳定站立
+        finished_count = 0
+        for _ in range(50):
+            finished = self.Player.skills_manager.execute("Neutral")
+            self.sync()
+            if finished:
+                finished_count += 1
+                if finished_count >= 20:  # 假设需要连续20次完成才算成功
+                    break
+
+        if self.enable_reset_perturb and self.reset_joint_noise_rad > 0.0:
+            perturb_action = np.zeros(self.no_of_actions, dtype=np.float32)
+            # Perturb waist + lower body only (10:), keep head/arms stable.
+            perturb_action[10:] = np.random.uniform(
+                -self.reset_joint_noise_rad,
+                self.reset_joint_noise_rad,
+                size=(self.no_of_actions - 10,)
+            )
+
+            for _ in range(self.reset_perturb_steps):
+                target_joint_positions = (self.joint_nominal_position + perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+            for i in range(self.reset_recover_steps):
+                # Linearly fade perturbation to help policy start from near-neutral.
+                alpha = 1.0 - float(i + 1) / float(self.reset_recover_steps)
+                target_joint_positions = (self.joint_nominal_position + alpha * perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+        # memory variables
+        self.sync()  
+        self.initial_position = np.array(self.Player.world.global_position[:2])
+        self.previous_pos = self.initial_position.copy()  # Critical: set to actual position
+        self.act = np.zeros(self.no_of_actions, np.float32)
+        # Build target in the robot's current forward direction instead of fixed global +x.
+        heading_deg = float(r.global_orientation_euler[2])
+        forward_offset = MathOps.rotate_2d_vec(np.array([length1, 0.0]), heading_deg, is_rad=False)
+        point1 = self.initial_position + forward_offset
+        point2 = point1 + MathOps.rotate_2d_vec(np.array([length2, 0]), angle2, is_rad=False)
+        point3 = point2 + MathOps.rotate_2d_vec(np.array([length3, 0]), angle3, is_rad=False)
+        self.point_list = [point1]
+        self.target_position = self.point_list[self.waypoint_index]
+        self.initial_height = self.Player.world.global_position[2]
+
+        return self.observe(True), {}
+
+    def render(self, mode='human', close=False):
+        return
+
+    def compute_reward(self, previous_pos, current_pos, action):
+        height = float(self.Player.world.global_position[2])
+
+        orientation_quat_inv = R.from_quat(self.Player.robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+        tilt_mag = float(np.linalg.norm(projected_gravity[:2]))
+        ang_vel = np.deg2rad(self.Player.robot.gyroscope)
+        ang_vel_mag = float(np.linalg.norm(ang_vel))
+
+        is_fallen = height < 0.3
+        if is_fallen:
+            # remain = max(0, 800 - self.step_counter)
+            # return -8.0 - 0.01 * remain
+            return -1.0
+        
+
+
+        # # 目标方向
+        # to_target = self.target_position - current_pos
+        # dist_to_target = float(np.linalg.norm(to_target))
+        # if dist_to_target < 0.5:
+        #     return 15.0
+
+        # forward_dir = to_target / dist_to_target if dist_to_target > 0.1 else np.array([1.0, 0.0])
+        # delta_pos = current_pos - previous_pos
+        # forward_step = float(np.dot(delta_pos, forward_dir))
+        # lateral_step = float(np.linalg.norm(delta_pos - forward_dir * forward_step))
+
+        # 奖励项
+        # progress_reward = 2 * forward_step
+        # lateral_penalty = -0.1 * lateral_step
+        alive_bonus = 2.0
+
+        # action_penalty = -0.01 * float(np.linalg.norm(action))
+        smoothness_penalty = -0.01 * float(np.linalg.norm(action - self.last_action_for_reward))
+
+        posture_penalty = -0.3 * (tilt_mag)
+        ang_vel_penalty = -0.02 * ang_vel_mag
+
+        target_height = self.initial_height
+        height_error =  height - target_height
+        height_penalty = -0.5 * abs(height_error)  # 惩罚高度偏离，系数可调
+
+        # # 在 compute_reward 开头附近，添加高度变化率计算
+        # if not hasattr(self, 'last_height'):
+        #     self.last_height = height
+        #     self.last_height_time = self.step_counter  # 可选，用于时间间隔
+        # height_rate = height - self.last_height  # 正为上升，负为下降
+        # self.last_height = height
+
+        # 惩罚高度下降（负变化率）
+        # height_down_penalty = -5.0 * max(0, -height_rate)  # 系数可调，-height_rate 为正表示下降幅度
+
+        # # 在 compute_reward 中
+        # if self.step_counter > 50:
+        #     avg_prev_action = np.mean(self.prev_action_history, axis=0)
+        #     novelty = float(np.linalg.norm(action - avg_prev_action))
+        #     exploration_bonus = 0.05 * novelty
+        # else:
+        #     exploration_bonus = 0
+
+        # self.prev_action_history[self.history_idx] = action
+        # self.history_idx = (self.history_idx + 1) % 50
+
+
+        total = (
+            # progress_reward  + 
+                 alive_bonus + 
+                 # lateral_penalty +
+                 # action_penalty +
+                smoothness_penalty +
+                 posture_penalty
+                 + ang_vel_penalty
+                 + height_penalty
+                #  + exploration_bonus
+                # + height_down_penalty
+                 )
+        if time.time() - self.start_time >= 1200:
+            self.start_time = time.time()
+            print(
+                #   f"progress_reward:{progress_reward:.4f}",
+                #   f"lateral_penalty:{lateral_penalty:.4f}",
+                #   f"action_penalty:{action_penalty:.4f}"s, 
+                  f"height_penalty:{height_penalty:.4f}",
+                  f"smoothness_penalty:{smoothness_penalty:.4f},",
+                  f"posture_penalty:{posture_penalty:.4f}",
+                #   f"ang_vel_penalty:{ang_vel_penalty:.4f}",
+                #   f"height_down_penalty:{height_down_penalty:.4f}",
+                #   f"exploration_bonus:{exploration_bonus:.4f}"
+            )
+
+        return total
+
+
+
+    def step(self, action):
+
+        r = self.Player.robot
+        self.previous_action = action
+
+        self.target_joint_positions = (
+                # self.joint_nominal_position +
+                 self.scaling_factor * action
+        )
+        self.target_joint_positions *= self.train_sim_flip
+
+        for idx, target in enumerate(self.target_joint_positions):
+            r.set_motor_target_position(
+                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=40, kd=1.0
+            )
+
+        self.previous_action = action
+
+        self.sync()  # run simulation step
+        self.step_counter += 1
+
+        if self.enable_debug_joint_status and self.step_counter % self.debug_every_n_steps == 0:
+            self.debug_joint_status()
+
+        current_pos = np.array(self.Player.world.global_position[:2], dtype=np.float32)
+
+        # Compute reward based on movement from previous step
+        reward = self.compute_reward(self.previous_pos, current_pos, action)
+
+        # Update previous position
+        self.previous_pos = current_pos.copy()
+        self.last_action_for_reward = action.copy()
+
+        # Fall detection and penalty
+        is_fallen = self.Player.world.global_position[2] < 0.3
+
+        # terminal state: the robot is falling or timeout
+        terminated = is_fallen or self.step_counter > 800 or self.route_completed
+        truncated = False
+
+        return self.observe(), reward, terminated, truncated, {}
+
+
+class Train(Train_Base):
+    def __init__(self, script) -> None:
+        super().__init__(script)
+
+    def train(self, args):
+
+        # --------------------------------------- Learning parameters
+        n_envs = 20  # Reduced from 8 to decrease CPU/network pressure during init
+        if n_envs < 1:
+            raise ValueError("GYM_CPU_N_ENVS must be >= 1")
+        n_steps_per_env = 256  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        minibatch_size = 512  # should be a factor of (n_steps_per_env * n_envs)
+        total_steps = 30000000
+        learning_rate = 3e-4
+        folder_name = f'Walk_R{self.robot_type}'
+        model_path = f'./scripts/gyms/logs/{folder_name}/'
+
+        print(f"Model path: {model_path}")
+        print(f"Using {n_envs} parallel environments")
+
+        # --------------------------------------- Run algorithm
+        def init_env(i_env):
+            def thunk():
+                return WalkEnv(self.ip, self.server_p + i_env)
+
+            return thunk
+
+        server_log_dir = os.path.join(model_path, "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1)  # include 1 extra server for testing
+
+        # Wait for servers to start
+        print(f"Starting {n_envs + 1} rcssservermj servers...")
+        print("Servers started, creating environments...")
+
+        env = SubprocVecEnv([init_env(i) for i in range(n_envs)])
+        eval_env = SubprocVecEnv([init_env(n_envs)])
+
+        try:
+            # Custom policy network architecture
+            policy_kwargs = dict(
+                net_arch=dict(
+                    pi=[512, 256, 128],  # Policy network: 3 layers
+                    vf=[512, 256, 128]  # Value network: 3 layers
+                ),
+                activation_fn=__import__('torch.nn', fromlist=['ELU']).ELU,
+            )
+
+            if "model_file" in args:  # retrain
+                model = PPO.load(args["model_file"], env=env, device="cpu", n_envs=n_envs, n_steps=n_steps_per_env,
+                                 batch_size=minibatch_size, learning_rate=learning_rate)
+            else:  # train new model
+                model = PPO(
+                    "MlpPolicy",
+                    env=env,
+                    verbose=1,
+                    n_steps=n_steps_per_env,
+                    batch_size=minibatch_size,
+                    learning_rate=learning_rate,
+                    device="cpu",
+                    policy_kwargs=policy_kwargs,
+                    ent_coef=0.05,  # Entropy coefficient for exploration
+                    # clip_range=0.13,  # PPO clipping parameter
+                    gae_lambda=0.95,  # GAE lambda
+                    gamma=0.95 , # Discount factor
+                    target_kl=0.03,
+                    n_epochs=5
+                )
+
+            model_path = self.learn_model(model, total_steps, model_path, eval_env=eval_env,
+                                          eval_freq=n_steps_per_env * 10, save_freq=n_steps_per_env * 10,
+                                          backup_env_file=__file__)
+        except KeyboardInterrupt:
+            sleep(1)  # wait for child processes
+            print("\nctrl+c pressed, aborting...\n")
+            servers.kill()
+            return
+
+        env.close()
+        eval_env.close()
+        servers.kill()
+
+    def test(self, args):
+
+        # Uses different server and monitor ports
+        server_log_dir = os.path.join(args["folder_dir"], "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        server = Train_Server(self.server_p - 1, self.monitor_p, 1)
+        env = WalkEnv(self.ip, self.server_p - 1)
+        model = PPO.load(args["model_file"], env=env)
+
+        try:
+            self.export_model(args["model_file"], args["model_file"] + ".pkl",
+                              False)  # Export to pkl to create custom behavior
+            self.test_model(model, env, log_path=args["folder_dir"], model_path=args["folder_dir"])
+        except KeyboardInterrupt:
+            print()
+
+        env.close()
+        server.kill()
+
+
+if __name__ == "__main__":
+    from types import SimpleNamespace
+
+    # 创建默认参数
+    script_args = SimpleNamespace(
+        args=SimpleNamespace(
+            i='127.0.0.1',  # Server IP
+            p=3100,  # Server port
+            m=3200,  # Monitor port
+            r=0,  # Robot type
+            t='Gym',  # Team name
+            u=1  # Uniform number
+        )
+    )
+
+    trainer = Train(script_args)
+    trainer.train({})
+    # trainer.test({"model_file": "scripts/gyms/logs/Walk_R0_000/best_model.zip",
+                #   "folder_dir": "scripts/gyms/logs/Walk_R0_000/",})

scripts/gyms/logs/Walk_R0_005/Walk.py

@@ -0,0 +1,660 @@
+import os
+import numpy as np
+import math
+import time
+from time import sleep
+from random import random
+from random import uniform
+
+from stable_baselines3 import PPO
+from stable_baselines3.common.monitor import Monitor
+from stable_baselines3.common.vec_env import SubprocVecEnv, DummyVecEnv
+
+import gymnasium as gym
+from gymnasium import spaces
+
+from scripts.commons.Train_Base import Train_Base
+from scripts.commons.Server import Server as Train_Server
+
+from agent.base_agent import Base_Agent
+from utils.math_ops import MathOps
+
+from scipy.spatial.transform import Rotation as R
+
+'''
+Objective:
+Learn how to run forward using step primitive
+----------
+- class Basic_Run: implements an OpenAI custom gym
+- class Train:  implements algorithms to train a new model or test an existing model
+'''
+
+
+class WalkEnv(gym.Env):
+    def __init__(self, ip, server_p) -> None:
+
+        # Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
+        self.Player = player = Base_Agent(
+            team_name="Gym",
+            number=1,
+            host=ip,
+            port=server_p
+        )
+        self.robot_type = self.Player.robot
+        self.step_counter = 0  # to limit episode size
+        self.force_play_on = True
+
+        self.target_position = np.array([0.0, 0.0])  # target position in the x-y plane
+        self.initial_position = np.array([0.0, 0.0])  # initial position in the x-y plane
+        self.target_direction = 0.0  # target direction in the x-y plane (relative to the robot's orientation)
+        self.isfallen = False
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.debug_every_n_steps = 5
+        self.enable_debug_joint_status = False
+        self.calibrate_nominal_from_neutral = True
+        self.auto_calibrate_train_sim_flip = True
+        self.nominal_calibrated_once = False
+        self.flip_calibrated_once = False
+        self._target_hz = 0.0
+        self._target_dt = 0.0
+        self._last_sync_time = None
+        target_hz_env = 0
+        if target_hz_env:
+            try:
+                self._target_hz = float(target_hz_env)
+            except ValueError:
+                self._target_hz = 0.0
+        if self._target_hz > 0.0:
+            self._target_dt = 1.0 / self._target_hz
+
+        # State space
+        # 原始观测大小: 78
+        obs_size = 78
+        self.obs = np.zeros(obs_size, np.float32)
+        self.observation_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(obs_size,),
+            dtype=np.float32
+        )
+
+        action_dim = len(self.Player.robot.ROBOT_MOTORS)
+        self.no_of_actions = action_dim
+        self.action_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(action_dim,),
+            dtype=np.float32
+        )
+
+        # 中立姿态
+        self.joint_nominal_position = np.array(
+            [
+                0.0,
+                0.0,
+                0.0,
+                1.4,
+                0.0,
+                -0.4,
+                0.0,
+                -1.4,
+                0.0,
+                0.4,
+                0.0,
+                -0.4,
+                0.0,
+                0.0,
+                0.8,
+                -0.4,
+                0.0,
+                0.4,
+                0.0,
+                0.0,
+                -0.8,
+                0.4,
+                0.0,
+            ]
+        )
+        self.joint_nominal_position = np.zeros(self.no_of_actions)
+        self.train_sim_flip = np.array(
+            [
+                1.0,  # 0: Head_yaw (he1)
+                -1.0,  # 1: Head_pitch (he2)
+                1.0,  # 2: Left_Shoulder_Pitch (lae1)
+                -1.0,  # 3: Left_Shoulder_Roll (lae2)
+                -1.0,  # 4: Left_Elbow_Pitch (lae3)
+                1.0,  # 5: Left_Elbow_Yaw (lae4)
+                -1.0,  # 6: Right_Shoulder_Pitch (rae1)
+                -1.0,  # 7: Right_Shoulder_Roll (rae2)
+                1.0,  # 8: Right_Elbow_Pitch (rae3)
+                1.0,  # 9: Right_Elbow_Yaw (rae4)
+                1.0,  # 10: Waist (te1)
+                1.0,  # 11: Left_Hip_Pitch (lle1)
+                -1.0,  # 12: Left_Hip_Roll (lle2)
+                -1.0,  # 13: Left_Hip_Yaw (lle3)
+                1.0,  # 14: Left_Knee_Pitch (lle4)
+                1.0,  # 15: Left_Ankle_Pitch (lle5)
+                -1.0,  # 16: Left_Ankle_Roll (lle6)
+                -1.0,  # 17: Right_Hip_Pitch (rle1)
+                -1.0,  # 18: Right_Hip_Roll (rle2)
+                -1.0,  # 19: Right_Hip_Yaw (rle3)
+                -1.0,  # 20: Right_Knee_Pitch (rle4)
+                -1.0,  # 21: Right_Ankle_Pitch (rle5)
+                -1.0,  # 22: Right_Ankle_Roll (rle6)
+            ]
+        )
+
+        self.scaling_factor = 0.3
+        # self.scaling_factor = 1
+
+        # Small reset perturbations for robustness training.
+        self.enable_reset_perturb = True
+        self.reset_beam_yaw_range_deg = 180  # randomize target direction fully to encourage learning a real walk instead of a fixed gait
+        self.reset_joint_noise_rad = 0.015
+        self.reset_perturb_steps = 3
+        self.reset_recover_steps = 8
+
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Track previous position
+        self.Player.server.connect()
+        # sleep(2.0)  # Longer wait for connection to establish completely
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+        self.start_time = time.time()
+
+    def _reconnect_server(self):
+        try:
+            self.Player.server.shutdown()
+        except Exception:
+            pass
+
+        self.Player.server.connect()
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+
+    def _safe_receive_world_update(self, retries=1):
+        last_exc = None
+        for attempt in range(retries + 1):
+            try:
+                self.Player.server.receive()
+                self.Player.world.update()
+                return
+            except (ConnectionResetError, OSError) as exc:
+                last_exc = exc
+                if attempt >= retries:
+                    raise
+                self._reconnect_server()
+        if last_exc is not None:
+            raise last_exc
+
+    def debug_log(self, message):
+        print(message)
+        try:
+            log_path = os.path.join(os.path.dirname(os.path.dirname(__file__)), "comm_debug.log")
+            with open(log_path, "a", encoding="utf-8") as f:
+                f.write(message + "\n")
+        except OSError:
+            pass
+
+    def observe(self, init=False):
+
+        """获取当前观测值"""
+        robot = self.Player.robot
+        world = self.Player.world
+
+        # Safety check: ensure data is available
+
+        # 计算目标速度
+        raw_target = self.target_position - world.global_position[:2]
+        velocity = MathOps.rotate_2d_vec(
+            raw_target,
+            -robot.global_orientation_euler[2],
+            is_rad=False
+        )
+
+        # 计算相对方向
+        rel_orientation = MathOps.vector_angle(velocity) * 0.3
+        rel_orientation = np.clip(rel_orientation, -0.25, 0.25)
+
+        velocity = np.concatenate([velocity, np.array([rel_orientation])])
+        velocity[0] = np.clip(velocity[0], -0.5, 0.5)
+        velocity[1] = np.clip(velocity[1], -0.25, 0.25)
+
+        # 关节状态
+        radian_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        radian_joint_speeds = np.deg2rad(
+            [robot.motor_speeds[motor] for motor in robot.ROBOT_MOTORS]
+        )
+
+        qpos_qvel_previous_action = np.concatenate([
+            (radian_joint_positions * self.train_sim_flip - self.joint_nominal_position) / 4.6,
+            radian_joint_speeds / 110.0 * self.train_sim_flip,
+            self.previous_action / 10.0,
+        ])
+
+        # 角速度
+        ang_vel = np.clip(np.deg2rad(robot.gyroscope) / 50.0, -1.0, 1.0)
+
+        # 投影的重力方向
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+
+        # 组合观测
+        observation = np.concatenate([
+            qpos_qvel_previous_action,
+            ang_vel,
+            velocity,
+            projected_gravity,
+        ])
+
+        observation = np.clip(observation, -10.0, 10.0)
+        return observation.astype(np.float32)
+
+    def sync(self):
+        ''' Run a single simulation step '''
+        self._safe_receive_world_update(retries=1)
+        self.Player.robot.commit_motor_targets_pd()
+        self.Player.server.send()
+        if self._target_dt > 0.0:
+            now = time.time()
+            if self._last_sync_time is None:
+                self._last_sync_time = now
+                return
+            elapsed = now - self._last_sync_time
+            remaining = self._target_dt - elapsed
+            if remaining > 0.0:
+                time.sleep(remaining)
+                now = time.time()
+            self._last_sync_time = now
+
+    def debug_joint_status(self):
+        robot = self.Player.robot
+        actual_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        target_joint_positions = getattr(
+            self,
+            'target_joint_positions',
+            np.zeros(len(robot.ROBOT_MOTORS), dtype=np.float32)
+        )
+        joint_error = actual_joint_positions - target_joint_positions
+        leg_slice = slice(11, None)
+
+        self.debug_log(
+            "[WalkDebug] "
+            f"step={self.step_counter} "
+            f"pos={np.round(self.Player.world.global_position, 3).tolist()} "
+            f"target_xy={np.round(self.target_position, 3).tolist()} "
+            f"target_leg={np.round(target_joint_positions[leg_slice], 3).tolist()} "
+            f"actual_leg={np.round(actual_joint_positions[leg_slice], 3).tolist()} "
+            f"err_norm={float(np.linalg.norm(joint_error)):.4f} "
+            f"fallen={self.Player.world.global_position[2] < 0.3}"
+        )
+        print(f"waist target={target_joint_positions[10]:.3f}, actual={actual_joint_positions[10]:.3f}")
+
+    def reset(self, seed=None, options=None):
+        '''
+        Reset and stabilize the robot
+        Note: for some behaviors it would be better to reduce stabilization or add noise
+        '''
+        r = self.Player.robot
+        super().reset(seed=seed)
+        if seed is not None:
+            np.random.seed(seed)
+
+        length1 = 2  # randomize target distance
+        length2 = np.random.uniform(0.6, 1)  # randomize target distance
+        length3 = np.random.uniform(0.6, 1)  # randomize target distance
+        angle2 = np.random.uniform(-30, 30)  # randomize initial orientation
+        angle3 = np.random.uniform(-30, 30)  # randomize target direction
+
+        self.step_counter = 0
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Initialize for first step
+        self.walk_cycle_step = 0
+
+        # 随机 beam 目标位置和朝向，增加训练多样性
+        beam_x = (random() - 0.5) * 10
+        beam_y = (random() - 0.5) * 10
+        beam_yaw = uniform(-self.reset_beam_yaw_range_deg, self.reset_beam_yaw_range_deg)
+
+        for _ in range(5):
+            self._safe_receive_world_update(retries=2)
+            self.Player.robot.commit_motor_targets_pd()
+            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=beam_yaw)
+            self.Player.server.send()
+
+        # 执行 Neutral 技能直到完成，给机器人足够时间在 beam 位置稳定站立
+        finished_count = 0
+        for _ in range(50):
+            finished = self.Player.skills_manager.execute("Neutral")
+            self.sync()
+            if finished:
+                finished_count += 1
+                if finished_count >= 20:  # 假设需要连续20次完成才算成功
+                    break
+
+        if self.enable_reset_perturb and self.reset_joint_noise_rad > 0.0:
+            perturb_action = np.zeros(self.no_of_actions, dtype=np.float32)
+            # Perturb waist + lower body only (10:), keep head/arms stable.
+            perturb_action[10:] = np.random.uniform(
+                -self.reset_joint_noise_rad,
+                self.reset_joint_noise_rad,
+                size=(self.no_of_actions - 10,)
+            )
+
+            for _ in range(self.reset_perturb_steps):
+                target_joint_positions = (self.joint_nominal_position + perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+            for i in range(self.reset_recover_steps):
+                # Linearly fade perturbation to help policy start from near-neutral.
+                alpha = 1.0 - float(i + 1) / float(self.reset_recover_steps)
+                target_joint_positions = (self.joint_nominal_position + alpha * perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+        # memory variables
+        self.sync()  
+        self.initial_position = np.array(self.Player.world.global_position[:2])
+        self.previous_pos = self.initial_position.copy()  # Critical: set to actual position
+        self.act = np.zeros(self.no_of_actions, np.float32)
+        # Build target in the robot's current forward direction instead of fixed global +x.
+        heading_deg = float(r.global_orientation_euler[2])
+        forward_offset = MathOps.rotate_2d_vec(np.array([length1, 0.0]), heading_deg, is_rad=False)
+        point1 = self.initial_position + forward_offset
+        point2 = point1 + MathOps.rotate_2d_vec(np.array([length2, 0]), angle2, is_rad=False)
+        point3 = point2 + MathOps.rotate_2d_vec(np.array([length3, 0]), angle3, is_rad=False)
+        self.point_list = [point1]
+        self.target_position = self.point_list[self.waypoint_index]
+        self.initial_height = self.Player.world.global_position[2]
+
+        return self.observe(True), {}
+
+    def render(self, mode='human', close=False):
+        return
+
+    def compute_reward(self, previous_pos, current_pos, action):
+        height = float(self.Player.world.global_position[2])
+
+        orientation_quat_inv = R.from_quat(self.Player.robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+        tilt_mag = float(np.linalg.norm(projected_gravity[:2]))
+        ang_vel = np.deg2rad(self.Player.robot.gyroscope)
+        ang_vel_mag = float(np.linalg.norm(ang_vel))
+
+        is_fallen = height < 0.3
+        if is_fallen:
+            # remain = max(0, 800 - self.step_counter)
+            # return -8.0 - 0.01 * remain
+            return -1.0
+        
+
+
+        # # 目标方向
+        # to_target = self.target_position - current_pos
+        # dist_to_target = float(np.linalg.norm(to_target))
+        # if dist_to_target < 0.5:
+        #     return 15.0
+
+        # forward_dir = to_target / dist_to_target if dist_to_target > 0.1 else np.array([1.0, 0.0])
+        # delta_pos = current_pos - previous_pos
+        # forward_step = float(np.dot(delta_pos, forward_dir))
+        # lateral_step = float(np.linalg.norm(delta_pos - forward_dir * forward_step))
+
+        # 奖励项
+        # progress_reward = 2 * forward_step
+        # lateral_penalty = -0.1 * lateral_step
+        alive_bonus = 2.0
+
+        # action_penalty = -0.01 * float(np.linalg.norm(action))
+        smoothness_penalty = -0.01 * float(np.linalg.norm(action - self.last_action_for_reward))
+
+        posture_penalty = -0.3 * (tilt_mag)
+        ang_vel_penalty = -0.02 * ang_vel_mag
+
+        target_height = self.initial_height
+        height_error =  height - target_height
+        height_penalty = -0.5 * abs(height_error)  # 惩罚高度偏离，系数可调
+
+        # # 在 compute_reward 开头附近，添加高度变化率计算
+        # if not hasattr(self, 'last_height'):
+        #     self.last_height = height
+        #     self.last_height_time = self.step_counter  # 可选，用于时间间隔
+        # height_rate = height - self.last_height  # 正为上升，负为下降
+        # self.last_height = height
+
+        # 惩罚高度下降（负变化率）
+        # height_down_penalty = -5.0 * max(0, -height_rate)  # 系数可调，-height_rate 为正表示下降幅度
+
+        # # 在 compute_reward 中
+        # if self.step_counter > 50:
+        #     avg_prev_action = np.mean(self.prev_action_history, axis=0)
+        #     novelty = float(np.linalg.norm(action - avg_prev_action))
+        #     exploration_bonus = 0.05 * novelty
+        # else:
+        #     exploration_bonus = 0
+
+        # self.prev_action_history[self.history_idx] = action
+        # self.history_idx = (self.history_idx + 1) % 50
+
+
+        total = (
+            # progress_reward  + 
+                 alive_bonus + 
+                 # lateral_penalty +
+                 # action_penalty +
+                smoothness_penalty +
+                 posture_penalty
+                 + ang_vel_penalty
+                 + height_penalty
+                #  + exploration_bonus
+                # + height_down_penalty
+                 )
+        if time.time() - self.start_time >= 1200:
+            self.start_time = time.time()
+            print(
+                #   f"progress_reward:{progress_reward:.4f}",
+                #   f"lateral_penalty:{lateral_penalty:.4f}",
+                #   f"action_penalty:{action_penalty:.4f}"s, 
+                  f"height_penalty:{height_penalty:.4f}",
+                  f"smoothness_penalty:{smoothness_penalty:.4f},",
+                  f"posture_penalty:{posture_penalty:.4f}",
+                #   f"ang_vel_penalty:{ang_vel_penalty:.4f}",
+                #   f"height_down_penalty:{height_down_penalty:.4f}",
+                #   f"exploration_bonus:{exploration_bonus:.4f}"
+            )
+
+        return total
+
+
+
+    def step(self, action):
+
+        r = self.Player.robot
+        self.previous_action = action
+
+        self.target_joint_positions = (
+                # self.joint_nominal_position +
+                 self.scaling_factor * action
+        )
+        self.target_joint_positions *= self.train_sim_flip
+
+        for idx, target in enumerate(self.target_joint_positions):
+            r.set_motor_target_position(
+                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=40, kd=1.0
+            )
+
+        self.previous_action = action
+
+        self.sync()  # run simulation step
+        self.step_counter += 1
+
+        if self.enable_debug_joint_status and self.step_counter % self.debug_every_n_steps == 0:
+            self.debug_joint_status()
+
+        current_pos = np.array(self.Player.world.global_position[:2], dtype=np.float32)
+
+        # Compute reward based on movement from previous step
+        reward = self.compute_reward(self.previous_pos, current_pos, action)
+
+        # Update previous position
+        self.previous_pos = current_pos.copy()
+        self.last_action_for_reward = action.copy()
+
+        # Fall detection and penalty
+        is_fallen = self.Player.world.global_position[2] < 0.3
+
+        # terminal state: the robot is falling or timeout
+        terminated = is_fallen or self.step_counter > 800 or self.route_completed
+        truncated = False
+
+        return self.observe(), reward, terminated, truncated, {}
+
+
+class Train(Train_Base):
+    def __init__(self, script) -> None:
+        super().__init__(script)
+
+    def train(self, args):
+
+        # --------------------------------------- Learning parameters
+        n_envs = int(os.environ.get("GYM_CPU_N_ENVS", "20"))
+        if n_envs < 1:
+            raise ValueError("GYM_CPU_N_ENVS must be >= 1")
+        server_warmup_sec = float(os.environ.get("GYM_CPU_SERVER_WARMUP_SEC", "3.0"))
+        n_steps_per_env = 256  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        minibatch_size = 512  # should be a factor of (n_steps_per_env * n_envs)
+        total_steps = 30000000
+        learning_rate = 1e-4
+        folder_name = f'Walk_R{self.robot_type}'
+        model_path = f'./scripts/gyms/logs/{folder_name}/'
+
+        print(f"Model path: {model_path}")
+        print(f"Using {n_envs} parallel environments")
+
+        # --------------------------------------- Run algorithm
+        def init_env(i_env, monitor=False):
+            def thunk():
+                env = WalkEnv(self.ip, self.server_p + i_env)
+                if monitor:
+                    env = Monitor(env)
+                return env
+
+            return thunk
+
+        server_log_dir = os.path.join(model_path, "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1)  # include 1 extra server for testing
+
+        # Wait for servers to start
+        print(f"Starting {n_envs + 1} rcssservermj servers...")
+        if server_warmup_sec > 0:
+            print(f"Waiting {server_warmup_sec:.1f}s for server warmup...")
+            sleep(server_warmup_sec)
+        print("Servers started, creating environments...")
+
+        env = SubprocVecEnv([init_env(i, monitor=True) for i in range(n_envs)])
+        # Use single-process eval env to avoid extra subprocess fragility during callback evaluation.
+        eval_env = DummyVecEnv([init_env(n_envs, monitor=True)])
+
+        try:
+            # Custom policy network architecture
+            policy_kwargs = dict(
+                net_arch=dict(
+                    pi=[512, 256, 128],  # Policy network: 3 layers
+                    vf=[512, 256, 128]  # Value network: 3 layers
+                ),
+                activation_fn=__import__('torch.nn', fromlist=['ELU']).ELU,
+            )
+
+            if "model_file" in args:  # retrain
+                model = PPO.load(args["model_file"], env=env, device="cpu", n_envs=n_envs, n_steps=n_steps_per_env,
+                                 batch_size=minibatch_size, learning_rate=learning_rate)
+            else:  # train new model
+                model = PPO(
+                    "MlpPolicy",
+                    env=env,
+                    verbose=1,
+                    n_steps=n_steps_per_env,
+                    batch_size=minibatch_size,
+                    learning_rate=learning_rate,
+                    device="cpu",
+                    policy_kwargs=policy_kwargs,
+                    ent_coef=0.03,  # Entropy coefficient for exploration
+                    clip_range=0.13,  # PPO clipping parameter
+                    gae_lambda=0.95,  # GAE lambda
+                    gamma=0.95 , # Discount factor
+                    target_kl=0.03,
+                    n_epochs=5,
+                    tensorboard_log=f"./scripts/gyms/logs/{folder_name}/tensorboard/"
+                )
+
+            model_path = self.learn_model(model, total_steps, model_path, eval_env=eval_env,
+                                          eval_freq=n_steps_per_env * 10, save_freq=n_steps_per_env * 10,
+                                          backup_env_file=__file__)
+        except KeyboardInterrupt:
+            sleep(1)  # wait for child processes
+            print("\nctrl+c pressed, aborting...\n")
+            servers.kill()
+            return
+
+        env.close()
+        eval_env.close()
+        servers.kill()
+
+    def test(self, args):
+
+        # Uses different server and monitor ports
+        server_log_dir = os.path.join(args["folder_dir"], "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        server = Train_Server(self.server_p - 1, self.monitor_p, 1)
+        env = WalkEnv(self.ip, self.server_p - 1)
+        model = PPO.load(args["model_file"], env=env)
+
+        try:
+            self.export_model(args["model_file"], args["model_file"] + ".pkl",
+                              False)  # Export to pkl to create custom behavior
+            self.test_model(model, env, log_path=args["folder_dir"], model_path=args["folder_dir"])
+        except KeyboardInterrupt:
+            print()
+
+        env.close()
+        server.kill()
+
+
+if __name__ == "__main__":
+    from types import SimpleNamespace
+
+    # 创建默认参数
+    script_args = SimpleNamespace(
+        args=SimpleNamespace(
+            i='127.0.0.1',  # Server IP
+            p=3100,  # Server port
+            m=3200,  # Monitor port
+            r=0,  # Robot type
+            t='Gym',  # Team name
+            u=1  # Uniform number
+        )
+    )
+
+    trainer = Train(script_args)
+    trainer.train({"model_file": "scripts/gyms/logs/Walk_R0_004/best_model.zip"})
+    # trainer.test({"model_file": "scripts/gyms/logs/Walk_R0_004/best_model.zip",
+                #   "folder_dir": "scripts/gyms/logs/Walk_R0_004/",})

scripts/gyms/logs/Walk_R0_006/Walk.py

@@ -0,0 +1,679 @@
+import os
+import numpy as np
+import math
+import time
+from time import sleep
+from random import random
+from random import uniform
+
+from stable_baselines3 import PPO
+from stable_baselines3.common.monitor import Monitor
+from stable_baselines3.common.vec_env import SubprocVecEnv, DummyVecEnv
+
+import gymnasium as gym
+from gymnasium import spaces
+
+from scripts.commons.Train_Base import Train_Base
+from scripts.commons.Server import Server as Train_Server
+
+from agent.base_agent import Base_Agent
+from utils.math_ops import MathOps
+
+from scipy.spatial.transform import Rotation as R
+
+'''
+Objective:
+Learn how to run forward using step primitive
+----------
+- class Basic_Run: implements an OpenAI custom gym
+- class Train:  implements algorithms to train a new model or test an existing model
+'''
+
+
+class WalkEnv(gym.Env):
+    def __init__(self, ip, server_p) -> None:
+
+        # Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
+        self.Player = player = Base_Agent(
+            team_name="Gym",
+            number=1,
+            host=ip,
+            port=server_p
+        )
+        self.robot_type = self.Player.robot
+        self.step_counter = 0  # to limit episode size
+        self.force_play_on = True
+
+        self.target_position = np.array([0.0, 0.0])  # target position in the x-y plane
+        self.initial_position = np.array([0.0, 0.0])  # initial position in the x-y plane
+        self.target_direction = 0.0  # target direction in the x-y plane (relative to the robot's orientation)
+        self.isfallen = False
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.debug_every_n_steps = 5
+        self.enable_debug_joint_status = False
+        self.calibrate_nominal_from_neutral = True
+        self.auto_calibrate_train_sim_flip = True
+        self.nominal_calibrated_once = False
+        self.flip_calibrated_once = False
+        self._target_hz = 0.0
+        self._target_dt = 0.0
+        self._last_sync_time = None
+        target_hz_env = 0
+        if target_hz_env:
+            try:
+                self._target_hz = float(target_hz_env)
+            except ValueError:
+                self._target_hz = 0.0
+        if self._target_hz > 0.0:
+            self._target_dt = 1.0 / self._target_hz
+
+        # State space
+        # 原始观测大小: 78
+        obs_size = 78
+        self.obs = np.zeros(obs_size, np.float32)
+        self.observation_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(obs_size,),
+            dtype=np.float32
+        )
+
+        action_dim = len(self.Player.robot.ROBOT_MOTORS)
+        self.no_of_actions = action_dim
+        self.action_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(action_dim,),
+            dtype=np.float32
+        )
+
+        # 中立姿态
+        self.joint_nominal_position = np.array(
+            [
+                0.0,
+                0.0,
+                0.0,
+                1.4,
+                0.0,
+                -0.4,
+                0.0,
+                -1.4,
+                0.0,
+                0.4,
+                0.0,
+                -0.4,
+                0.0,
+                0.0,
+                0.8,
+                -0.4,
+                0.0,
+                0.4,
+                0.0,
+                0.0,
+                -0.8,
+                0.4,
+                0.0,
+            ]
+        )
+        self.joint_nominal_position = np.zeros(self.no_of_actions)
+        self.train_sim_flip = np.array(
+            [
+                1.0,  # 0: Head_yaw (he1)
+                -1.0,  # 1: Head_pitch (he2)
+                1.0,  # 2: Left_Shoulder_Pitch (lae1)
+                -1.0,  # 3: Left_Shoulder_Roll (lae2)
+                -1.0,  # 4: Left_Elbow_Pitch (lae3)
+                1.0,  # 5: Left_Elbow_Yaw (lae4)
+                -1.0,  # 6: Right_Shoulder_Pitch (rae1)
+                -1.0,  # 7: Right_Shoulder_Roll (rae2)
+                1.0,  # 8: Right_Elbow_Pitch (rae3)
+                1.0,  # 9: Right_Elbow_Yaw (rae4)
+                1.0,  # 10: Waist (te1)
+                1.0,  # 11: Left_Hip_Pitch (lle1)
+                -1.0,  # 12: Left_Hip_Roll (lle2)
+                -1.0,  # 13: Left_Hip_Yaw (lle3)
+                1.0,  # 14: Left_Knee_Pitch (lle4)
+                1.0,  # 15: Left_Ankle_Pitch (lle5)
+                -1.0,  # 16: Left_Ankle_Roll (lle6)
+                -1.0,  # 17: Right_Hip_Pitch (rle1)
+                -1.0,  # 18: Right_Hip_Roll (rle2)
+                -1.0,  # 19: Right_Hip_Yaw (rle3)
+                -1.0,  # 20: Right_Knee_Pitch (rle4)
+                -1.0,  # 21: Right_Ankle_Pitch (rle5)
+                -1.0,  # 22: Right_Ankle_Roll (rle6)
+            ]
+        )
+
+        self.scaling_factor = 0.3
+        # self.scaling_factor = 1
+
+        # Small reset perturbations for robustness training.
+        self.enable_reset_perturb = True
+        self.reset_beam_yaw_range_deg = 180  # randomize target direction fully to encourage learning a real walk instead of a fixed gait
+        self.reset_joint_noise_rad = 0.015
+        self.reset_perturb_steps = 3
+        self.reset_recover_steps = 8
+
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Track previous position
+        self.Player.server.connect()
+        # sleep(2.0)  # Longer wait for connection to establish completely
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+        self.start_time = time.time()
+
+    def _reconnect_server(self):
+        try:
+            self.Player.server.shutdown()
+        except Exception:
+            pass
+
+        self.Player.server.connect()
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+
+    def _safe_receive_world_update(self, retries=1):
+        last_exc = None
+        for attempt in range(retries + 1):
+            try:
+                self.Player.server.receive()
+                self.Player.world.update()
+                return
+            except (ConnectionResetError, OSError) as exc:
+                last_exc = exc
+                if attempt >= retries:
+                    raise
+                self._reconnect_server()
+        if last_exc is not None:
+            raise last_exc
+
+    def debug_log(self, message):
+        print(message)
+        try:
+            log_path = os.path.join(os.path.dirname(os.path.dirname(__file__)), "comm_debug.log")
+            with open(log_path, "a", encoding="utf-8") as f:
+                f.write(message + "\n")
+        except OSError:
+            pass
+
+    def observe(self, init=False):
+
+        """获取当前观测值"""
+        robot = self.Player.robot
+        world = self.Player.world
+
+        # Safety check: ensure data is available
+
+        # 计算目标速度
+        raw_target = self.target_position - world.global_position[:2]
+        velocity = MathOps.rotate_2d_vec(
+            raw_target,
+            -robot.global_orientation_euler[2],
+            is_rad=False
+        )
+
+        # 计算相对方向
+        rel_orientation = MathOps.vector_angle(velocity) * 0.3
+        rel_orientation = np.clip(rel_orientation, -0.25, 0.25)
+
+        velocity = np.concatenate([velocity, np.array([rel_orientation])])
+        velocity[0] = np.clip(velocity[0], -0.5, 0.5)
+        velocity[1] = np.clip(velocity[1], -0.25, 0.25)
+
+        # 关节状态
+        radian_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        radian_joint_speeds = np.deg2rad(
+            [robot.motor_speeds[motor] for motor in robot.ROBOT_MOTORS]
+        )
+
+        qpos_qvel_previous_action = np.concatenate([
+            (radian_joint_positions * self.train_sim_flip - self.joint_nominal_position) / 4.6,
+            radian_joint_speeds / 110.0 * self.train_sim_flip,
+            self.previous_action / 10.0,
+        ])
+
+        # 角速度
+        ang_vel = np.clip(np.deg2rad(robot.gyroscope) / 50.0, -1.0, 1.0)
+
+        # 投影的重力方向
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+
+        # 组合观测
+        observation = np.concatenate([
+            qpos_qvel_previous_action,
+            ang_vel,
+            velocity,
+            projected_gravity,
+        ])
+
+        observation = np.clip(observation, -10.0, 10.0)
+        return observation.astype(np.float32)
+
+    def sync(self):
+        ''' Run a single simulation step '''
+        self._safe_receive_world_update(retries=1)
+        self.Player.robot.commit_motor_targets_pd()
+        self.Player.server.send()
+        if self._target_dt > 0.0:
+            now = time.time()
+            if self._last_sync_time is None:
+                self._last_sync_time = now
+                return
+            elapsed = now - self._last_sync_time
+            remaining = self._target_dt - elapsed
+            if remaining > 0.0:
+                time.sleep(remaining)
+                now = time.time()
+            self._last_sync_time = now
+
+    def debug_joint_status(self):
+        robot = self.Player.robot
+        actual_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        target_joint_positions = getattr(
+            self,
+            'target_joint_positions',
+            np.zeros(len(robot.ROBOT_MOTORS), dtype=np.float32)
+        )
+        joint_error = actual_joint_positions - target_joint_positions
+        leg_slice = slice(11, None)
+
+        self.debug_log(
+            "[WalkDebug] "
+            f"step={self.step_counter} "
+            f"pos={np.round(self.Player.world.global_position, 3).tolist()} "
+            f"target_xy={np.round(self.target_position, 3).tolist()} "
+            f"target_leg={np.round(target_joint_positions[leg_slice], 3).tolist()} "
+            f"actual_leg={np.round(actual_joint_positions[leg_slice], 3).tolist()} "
+            f"err_norm={float(np.linalg.norm(joint_error)):.4f} "
+            f"fallen={self.Player.world.global_position[2] < 0.3}"
+        )
+        print(f"waist target={target_joint_positions[10]:.3f}, actual={actual_joint_positions[10]:.3f}")
+
+    def reset(self, seed=None, options=None):
+        '''
+        Reset and stabilize the robot
+        Note: for some behaviors it would be better to reduce stabilization or add noise
+        '''
+        r = self.Player.robot
+        super().reset(seed=seed)
+        if seed is not None:
+            np.random.seed(seed)
+
+        length1 = 2  # randomize target distance
+        length2 = np.random.uniform(0.6, 1)  # randomize target distance
+        length3 = np.random.uniform(0.6, 1)  # randomize target distance
+        angle2 = np.random.uniform(-30, 30)  # randomize initial orientation
+        angle3 = np.random.uniform(-30, 30)  # randomize target direction
+
+        self.step_counter = 0
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Initialize for first step
+        self.walk_cycle_step = 0
+
+        # 随机 beam 目标位置和朝向，增加训练多样性
+        beam_x = (random() - 0.5) * 10
+        beam_y = (random() - 0.5) * 10
+        beam_yaw = uniform(-self.reset_beam_yaw_range_deg, self.reset_beam_yaw_range_deg)
+
+        for _ in range(5):
+            self._safe_receive_world_update(retries=2)
+            self.Player.robot.commit_motor_targets_pd()
+            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=beam_yaw)
+            self.Player.server.send()
+
+        # 执行 Neutral 技能直到完成，给机器人足够时间在 beam 位置稳定站立
+        finished_count = 0
+        for _ in range(50):
+            finished = self.Player.skills_manager.execute("Neutral")
+            self.sync()
+            if finished:
+                finished_count += 1
+                if finished_count >= 20:  # 假设需要连续20次完成才算成功
+                    break
+
+        if self.enable_reset_perturb and self.reset_joint_noise_rad > 0.0:
+            perturb_action = np.zeros(self.no_of_actions, dtype=np.float32)
+            # Perturb waist + lower body only (10:), keep head/arms stable.
+            perturb_action[10:] = np.random.uniform(
+                -self.reset_joint_noise_rad,
+                self.reset_joint_noise_rad,
+                size=(self.no_of_actions - 10,)
+            )
+
+            for _ in range(self.reset_perturb_steps):
+                target_joint_positions = (self.joint_nominal_position + perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+            for i in range(self.reset_recover_steps):
+                # Linearly fade perturbation to help policy start from near-neutral.
+                alpha = 1.0 - float(i + 1) / float(self.reset_recover_steps)
+                target_joint_positions = (self.joint_nominal_position + alpha * perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+        # memory variables
+        self.sync()  
+        self.initial_position = np.array(self.Player.world.global_position[:2])
+        self.previous_pos = self.initial_position.copy()  # Critical: set to actual position
+        self.act = np.zeros(self.no_of_actions, np.float32)
+        # Build target in the robot's current forward direction instead of fixed global +x.
+        heading_deg = float(r.global_orientation_euler[2])
+        forward_offset = MathOps.rotate_2d_vec(np.array([length1, 0.0]), heading_deg, is_rad=False)
+        point1 = self.initial_position + forward_offset
+        point2 = point1 + MathOps.rotate_2d_vec(np.array([length2, 0]), angle2, is_rad=False)
+        point3 = point2 + MathOps.rotate_2d_vec(np.array([length3, 0]), angle3, is_rad=False)
+        self.point_list = [point1]
+        self.target_position = self.point_list[self.waypoint_index]
+        self.initial_height = self.Player.world.global_position[2]
+
+        return self.observe(True), {}
+
+    def render(self, mode='human', close=False):
+        return
+
+    def compute_reward(self, previous_pos, current_pos, action):
+        height = float(self.Player.world.global_position[2])
+
+        orientation_quat_inv = R.from_quat(self.Player.robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+        tilt_mag = float(np.linalg.norm(projected_gravity[:2]))
+        ang_vel = np.deg2rad(self.Player.robot.gyroscope)
+        ang_vel_mag = float(np.linalg.norm(ang_vel))
+
+        is_fallen = height < 0.3
+        if is_fallen:
+            # remain = max(0, 800 - self.step_counter)
+            # return -8.0 - 0.01 * remain
+            return -1.0
+        
+
+
+        # # 目标方向
+        # to_target = self.target_position - current_pos
+        # dist_to_target = float(np.linalg.norm(to_target))
+        # if dist_to_target < 0.5:
+        #     return 15.0
+
+        # forward_dir = to_target / dist_to_target if dist_to_target > 0.1 else np.array([1.0, 0.0])
+        # delta_pos = current_pos - previous_pos
+        # forward_step = float(np.dot(delta_pos, forward_dir))
+        # lateral_step = float(np.linalg.norm(delta_pos - forward_dir * forward_step))
+
+        # 奖励项
+        # progress_reward = 2 * forward_step
+        # lateral_penalty = -0.1 * lateral_step
+        alive_bonus = 2.0
+
+        # action_penalty = -0.01 * float(np.linalg.norm(action))
+        smoothness_penalty = -0.01 * float(np.linalg.norm(action - self.last_action_for_reward))
+
+        posture_penalty = -0.3 * (tilt_mag)
+        ang_vel_penalty = -0.02 * ang_vel_mag
+
+        target_height = self.initial_height
+        height_error =  height - target_height
+        height_penalty = -0.5 * abs(height_error)  # 惩罚高度偏离，系数可调
+
+        # # 在 compute_reward 开头附近，添加高度变化率计算
+        # if not hasattr(self, 'last_height'):
+        #     self.last_height = height
+        #     self.last_height_time = self.step_counter  # 可选，用于时间间隔
+        # height_rate = height - self.last_height  # 正为上升，负为下降
+        # self.last_height = height
+
+        # 惩罚高度下降（负变化率）
+        # height_down_penalty = -5.0 * max(0, -height_rate)  # 系数可调，-height_rate 为正表示下降幅度
+
+        # # 在 compute_reward 中
+        # if self.step_counter > 50:
+        #     avg_prev_action = np.mean(self.prev_action_history, axis=0)
+        #     novelty = float(np.linalg.norm(action - avg_prev_action))
+        #     exploration_bonus = 0.05 * novelty
+        # else:
+        #     exploration_bonus = 0
+
+        # self.prev_action_history[self.history_idx] = action
+        # self.history_idx = (self.history_idx + 1) % 50
+
+
+        total = (
+            # progress_reward  + 
+                 alive_bonus + 
+                 # lateral_penalty +
+                 # action_penalty +
+                smoothness_penalty +
+                 posture_penalty
+                 + ang_vel_penalty
+                 + height_penalty
+                #  + exploration_bonus
+                # + height_down_penalty
+                 )
+        if time.time() - self.start_time >= 600:
+            self.start_time = time.time()
+            print(
+                #   f"progress_reward:{progress_reward:.4f}",
+                #   f"lateral_penalty:{lateral_penalty:.4f}",
+                #   f"action_penalty:{action_penalty:.4f}"s, 
+                  f"height_penalty:{height_penalty:.4f}",
+                  f"smoothness_penalty:{smoothness_penalty:.4f},",
+                  f"posture_penalty:{posture_penalty:.4f}",
+                #   f"ang_vel_penalty:{ang_vel_penalty:.4f}",
+                #   f"height_down_penalty:{height_down_penalty:.4f}",
+                #   f"exploration_bonus:{exploration_bonus:.4f}"
+            )
+
+        return total
+
+
+
+    def step(self, action):
+
+        r = self.Player.robot
+        self.previous_action = action
+
+        self.target_joint_positions = (
+                # self.joint_nominal_position +
+                 self.scaling_factor * action
+        )
+        self.target_joint_positions *= self.train_sim_flip
+
+        for idx, target in enumerate(self.target_joint_positions):
+            r.set_motor_target_position(
+                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=40, kd=1.0
+            )
+
+        self.previous_action = action
+
+        self.sync()  # run simulation step
+        self.step_counter += 1
+
+        if self.enable_debug_joint_status and self.step_counter % self.debug_every_n_steps == 0:
+            self.debug_joint_status()
+
+        current_pos = np.array(self.Player.world.global_position[:2], dtype=np.float32)
+
+        # Compute reward based on movement from previous step
+        reward = self.compute_reward(self.previous_pos, current_pos, action)
+
+        # Update previous position
+        self.previous_pos = current_pos.copy()
+        self.last_action_for_reward = action.copy()
+
+        # Fall detection and penalty
+        is_fallen = self.Player.world.global_position[2] < 0.3
+
+        # terminal state: the robot is falling or timeout
+        terminated = is_fallen or self.step_counter > 800 or self.route_completed
+        truncated = False
+
+        return self.observe(), reward, terminated, truncated, {}
+
+
+class Train(Train_Base):
+    def __init__(self, script) -> None:
+        super().__init__(script)
+
+    def train(self, args):
+
+        # --------------------------------------- Learning parameters
+        n_envs = int(os.environ.get("GYM_CPU_N_ENVS", "20"))
+        if n_envs < 1:
+            raise ValueError("GYM_CPU_N_ENVS must be >= 1")
+        server_warmup_sec = float(os.environ.get("GYM_CPU_SERVER_WARMUP_SEC", "3.0"))
+        n_steps_per_env = int(os.environ.get("GYM_CPU_TRAIN_STEPS_PER_ENV", "256"))  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        minibatch_size = int(os.environ.get("GYM_CPU_TRAIN_BATCH_SIZE", "512"))  # should be a factor of (n_steps_per_env * n_envs)
+        total_steps = 30000000
+        learning_rate = float(os.environ.get("GYM_CPU_TRAIN_LR", "3e-4"))
+        folder_name = f'Walk_R{self.robot_type}'
+        model_path = f'./scripts/gyms/logs/{folder_name}/'
+
+        print(f"Model path: {model_path}")
+        print(f"Using {n_envs} parallel environments")
+
+        # --------------------------------------- Run algorithm
+        def init_env(i_env, monitor=False):
+            def thunk():
+                env = WalkEnv(self.ip, self.server_p + i_env)
+                if monitor:
+                    env = Monitor(env)
+                return env
+
+            return thunk
+
+        server_log_dir = os.path.join(model_path, "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1, no_render=True, no_realtime=True)  # include 1 extra server for testing
+
+        # Wait for servers to start
+        print(f"Starting {n_envs + 1} rcssservermj servers...")
+        if server_warmup_sec > 0:
+            print(f"Waiting {server_warmup_sec:.1f}s for server warmup...")
+            sleep(server_warmup_sec)
+        print("Servers started, creating environments...")
+
+        env = SubprocVecEnv([init_env(i, monitor=True) for i in range(n_envs)])
+        # Use single-process eval env to avoid extra subprocess fragility during callback evaluation.
+        eval_env = DummyVecEnv([init_env(n_envs, monitor=True)])
+
+        try:
+            # Custom policy network architecture
+            policy_kwargs = dict(
+                net_arch=dict(
+                    pi=[512, 256, 128],  # Policy network: 3 layers
+                    vf=[512, 256, 128]  # Value network: 3 layers
+                ),
+                activation_fn=__import__('torch.nn', fromlist=['ELU']).ELU,
+            )
+
+            if "model_file" in args:  # retrain
+                model = PPO.load(args["model_file"], env=env, device="cpu", n_envs=n_envs, n_steps=n_steps_per_env,
+                                 batch_size=minibatch_size, learning_rate=learning_rate)
+            else:  # train new model
+                model = PPO(
+                    "MlpPolicy",
+                    env=env,
+                    verbose=1,
+                    n_steps=n_steps_per_env,
+                    batch_size=minibatch_size,
+                    learning_rate=learning_rate,
+                    device="cpu",
+                    policy_kwargs=policy_kwargs,
+                    ent_coef=float(os.environ.get("GYM_CPU_TRAIN_ENT_COEF", "0.05")),  # Entropy coefficient for exploration
+                    clip_range=float(os.environ.get("GYM_CPU_TRAIN_CLIP_RANGE", "0.2")),  # PPO clipping parameter
+                    gae_lambda=0.95,  # GAE lambda
+                    gamma=float(os.environ.get("GYM_CPU_TRAIN_GAMMA", "0.95")), # Discount factor
+                    target_kl=0.03,
+                    n_epochs=int(os.environ.get("GYM_CPU_TRAIN_EPOCHS", "5")),
+                    # tensorboard_log=f"./scripts/gyms/logs/{folder_name}/tensorboard/"
+                )
+
+            model_path = self.learn_model(model, total_steps, model_path, eval_env=eval_env,
+                                          eval_freq=n_steps_per_env * 20, save_freq=n_steps_per_env * 20,
+                                          backup_env_file=__file__)
+        except KeyboardInterrupt:
+            sleep(1)  # wait for child processes
+            print("\nctrl+c pressed, aborting...\n")
+            servers.kill()
+            return
+
+        env.close()
+        eval_env.close()
+        servers.kill()
+
+    def test(self, args):
+
+        # Uses different server and monitor ports
+        server_log_dir = os.path.join(args["folder_dir"], "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        test_no_render = os.environ.get("GYM_CPU_TEST_NO_RENDER", "0") == "1"
+        test_no_realtime = os.environ.get("GYM_CPU_TEST_NO_REALTIME", "0") == "1"
+
+        server = Train_Server(
+            self.server_p - 1,
+            self.monitor_p,
+            1,
+            no_render=test_no_render,
+            no_realtime=test_no_realtime,
+        )
+        env = WalkEnv(self.ip, self.server_p - 1)
+        model = PPO.load(args["model_file"], env=env)
+
+        try:
+            self.export_model(args["model_file"], args["model_file"] + ".pkl",
+                              False)  # Export to pkl to create custom behavior
+            self.test_model(model, env, log_path=args["folder_dir"], model_path=args["folder_dir"])
+        except KeyboardInterrupt:
+            print()
+
+        env.close()
+        server.kill()
+
+
+if __name__ == "__main__":
+    from types import SimpleNamespace
+
+    # 创建默认参数
+    script_args = SimpleNamespace(
+        args=SimpleNamespace(
+            i='127.0.0.1',  # Server IP
+            p=3100,  # Server port
+            m=3200,  # Monitor port
+            r=0,  # Robot type
+            t='Gym',  # Team name
+            u=1  # Uniform number
+        )
+    )
+
+    trainer = Train(script_args)
+
+    run_mode = os.environ.get("GYM_CPU_MODE", "train").strip().lower()
+
+    if run_mode == "test":
+        test_model_file = os.environ.get("GYM_CPU_TEST_MODEL", "scripts/gyms/logs/Walk_R0_004/best_model.zip")
+        test_folder = os.environ.get("GYM_CPU_TEST_FOLDER", "scripts/gyms/logs/Walk_R0_004/")
+        trainer.test({"model_file": test_model_file, "folder_dir": test_folder})
+    else:
+        retrain_model = os.environ.get("GYM_CPU_TRAIN_MODEL", "").strip()
+        if retrain_model:
+            trainer.train({"model_file": retrain_model})
+        else:
+            trainer.train({})

scripts/gyms/logs/Walk_R0_007/Walk.py

@@ -0,0 +1,679 @@
+import os
+import numpy as np
+import math
+import time
+from time import sleep
+from random import random
+from random import uniform
+
+from stable_baselines3 import PPO
+from stable_baselines3.common.monitor import Monitor
+from stable_baselines3.common.vec_env import SubprocVecEnv, DummyVecEnv
+
+import gymnasium as gym
+from gymnasium import spaces
+
+from scripts.commons.Train_Base import Train_Base
+from scripts.commons.Server import Server as Train_Server
+
+from agent.base_agent import Base_Agent
+from utils.math_ops import MathOps
+
+from scipy.spatial.transform import Rotation as R
+
+'''
+Objective:
+Learn how to run forward using step primitive
+----------
+- class Basic_Run: implements an OpenAI custom gym
+- class Train:  implements algorithms to train a new model or test an existing model
+'''
+
+
+class WalkEnv(gym.Env):
+    def __init__(self, ip, server_p) -> None:
+
+        # Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
+        self.Player = player = Base_Agent(
+            team_name="Gym",
+            number=1,
+            host=ip,
+            port=server_p
+        )
+        self.robot_type = self.Player.robot
+        self.step_counter = 0  # to limit episode size
+        self.force_play_on = True
+
+        self.target_position = np.array([0.0, 0.0])  # target position in the x-y plane
+        self.initial_position = np.array([0.0, 0.0])  # initial position in the x-y plane
+        self.target_direction = 0.0  # target direction in the x-y plane (relative to the robot's orientation)
+        self.isfallen = False
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.debug_every_n_steps = 5
+        self.enable_debug_joint_status = False
+        self.calibrate_nominal_from_neutral = True
+        self.auto_calibrate_train_sim_flip = True
+        self.nominal_calibrated_once = False
+        self.flip_calibrated_once = False
+        self._target_hz = 0.0
+        self._target_dt = 0.0
+        self._last_sync_time = None
+        target_hz_env = 0
+        if target_hz_env:
+            try:
+                self._target_hz = float(target_hz_env)
+            except ValueError:
+                self._target_hz = 0.0
+        if self._target_hz > 0.0:
+            self._target_dt = 1.0 / self._target_hz
+
+        # State space
+        # 原始观测大小: 78
+        obs_size = 78
+        self.obs = np.zeros(obs_size, np.float32)
+        self.observation_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(obs_size,),
+            dtype=np.float32
+        )
+
+        action_dim = len(self.Player.robot.ROBOT_MOTORS)
+        self.no_of_actions = action_dim
+        self.action_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(action_dim,),
+            dtype=np.float32
+        )
+
+        # 中立姿态
+        self.joint_nominal_position = np.array(
+            [
+                0.0,
+                0.0,
+                0.0,
+                1.4,
+                0.0,
+                -0.4,
+                0.0,
+                -1.4,
+                0.0,
+                0.4,
+                0.0,
+                -0.4,
+                0.0,
+                0.0,
+                0.8,
+                -0.4,
+                0.0,
+                0.4,
+                0.0,
+                0.0,
+                -0.8,
+                0.4,
+                0.0,
+            ]
+        )
+        self.joint_nominal_position = np.zeros(self.no_of_actions)
+        self.train_sim_flip = np.array(
+            [
+                1.0,  # 0: Head_yaw (he1)
+                -1.0,  # 1: Head_pitch (he2)
+                1.0,  # 2: Left_Shoulder_Pitch (lae1)
+                -1.0,  # 3: Left_Shoulder_Roll (lae2)
+                -1.0,  # 4: Left_Elbow_Pitch (lae3)
+                1.0,  # 5: Left_Elbow_Yaw (lae4)
+                -1.0,  # 6: Right_Shoulder_Pitch (rae1)
+                -1.0,  # 7: Right_Shoulder_Roll (rae2)
+                1.0,  # 8: Right_Elbow_Pitch (rae3)
+                1.0,  # 9: Right_Elbow_Yaw (rae4)
+                1.0,  # 10: Waist (te1)
+                1.0,  # 11: Left_Hip_Pitch (lle1)
+                -1.0,  # 12: Left_Hip_Roll (lle2)
+                -1.0,  # 13: Left_Hip_Yaw (lle3)
+                1.0,  # 14: Left_Knee_Pitch (lle4)
+                1.0,  # 15: Left_Ankle_Pitch (lle5)
+                -1.0,  # 16: Left_Ankle_Roll (lle6)
+                -1.0,  # 17: Right_Hip_Pitch (rle1)
+                -1.0,  # 18: Right_Hip_Roll (rle2)
+                -1.0,  # 19: Right_Hip_Yaw (rle3)
+                -1.0,  # 20: Right_Knee_Pitch (rle4)
+                -1.0,  # 21: Right_Ankle_Pitch (rle5)
+                -1.0,  # 22: Right_Ankle_Roll (rle6)
+            ]
+        )
+
+        self.scaling_factor = 0.3
+        # self.scaling_factor = 1
+
+        # Small reset perturbations for robustness training.
+        self.enable_reset_perturb = True
+        self.reset_beam_yaw_range_deg = 180  # randomize target direction fully to encourage learning a real walk instead of a fixed gait
+        self.reset_joint_noise_rad = 0.015
+        self.reset_perturb_steps = 3
+        self.reset_recover_steps = 8
+
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Track previous position
+        self.Player.server.connect()
+        # sleep(2.0)  # Longer wait for connection to establish completely
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+        self.start_time = time.time()
+
+    def _reconnect_server(self):
+        try:
+            self.Player.server.shutdown()
+        except Exception:
+            pass
+
+        self.Player.server.connect()
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+
+    def _safe_receive_world_update(self, retries=1):
+        last_exc = None
+        for attempt in range(retries + 1):
+            try:
+                self.Player.server.receive()
+                self.Player.world.update()
+                return
+            except (ConnectionResetError, OSError) as exc:
+                last_exc = exc
+                if attempt >= retries:
+                    raise
+                self._reconnect_server()
+        if last_exc is not None:
+            raise last_exc
+
+    def debug_log(self, message):
+        print(message)
+        try:
+            log_path = os.path.join(os.path.dirname(os.path.dirname(__file__)), "comm_debug.log")
+            with open(log_path, "a", encoding="utf-8") as f:
+                f.write(message + "\n")
+        except OSError:
+            pass
+
+    def observe(self, init=False):
+
+        """获取当前观测值"""
+        robot = self.Player.robot
+        world = self.Player.world
+
+        # Safety check: ensure data is available
+
+        # 计算目标速度
+        raw_target = self.target_position - world.global_position[:2]
+        velocity = MathOps.rotate_2d_vec(
+            raw_target,
+            -robot.global_orientation_euler[2],
+            is_rad=False
+        )
+
+        # 计算相对方向
+        rel_orientation = MathOps.vector_angle(velocity) * 0.3
+        rel_orientation = np.clip(rel_orientation, -0.25, 0.25)
+
+        velocity = np.concatenate([velocity, np.array([rel_orientation])])
+        velocity[0] = np.clip(velocity[0], -0.5, 0.5)
+        velocity[1] = np.clip(velocity[1], -0.25, 0.25)
+
+        # 关节状态
+        radian_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        radian_joint_speeds = np.deg2rad(
+            [robot.motor_speeds[motor] for motor in robot.ROBOT_MOTORS]
+        )
+
+        qpos_qvel_previous_action = np.concatenate([
+            (radian_joint_positions * self.train_sim_flip - self.joint_nominal_position) / 4.6,
+            radian_joint_speeds / 110.0 * self.train_sim_flip,
+            self.previous_action / 10.0,
+        ])
+
+        # 角速度
+        ang_vel = np.clip(np.deg2rad(robot.gyroscope) / 50.0, -1.0, 1.0)
+
+        # 投影的重力方向
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+
+        # 组合观测
+        observation = np.concatenate([
+            qpos_qvel_previous_action,
+            ang_vel,
+            velocity,
+            projected_gravity,
+        ])
+
+        observation = np.clip(observation, -10.0, 10.0)
+        return observation.astype(np.float32)
+
+    def sync(self):
+        ''' Run a single simulation step '''
+        self._safe_receive_world_update(retries=1)
+        self.Player.robot.commit_motor_targets_pd()
+        self.Player.server.send()
+        if self._target_dt > 0.0:
+            now = time.time()
+            if self._last_sync_time is None:
+                self._last_sync_time = now
+                return
+            elapsed = now - self._last_sync_time
+            remaining = self._target_dt - elapsed
+            if remaining > 0.0:
+                time.sleep(remaining)
+                now = time.time()
+            self._last_sync_time = now
+
+    def debug_joint_status(self):
+        robot = self.Player.robot
+        actual_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        target_joint_positions = getattr(
+            self,
+            'target_joint_positions',
+            np.zeros(len(robot.ROBOT_MOTORS), dtype=np.float32)
+        )
+        joint_error = actual_joint_positions - target_joint_positions
+        leg_slice = slice(11, None)
+
+        self.debug_log(
+            "[WalkDebug] "
+            f"step={self.step_counter} "
+            f"pos={np.round(self.Player.world.global_position, 3).tolist()} "
+            f"target_xy={np.round(self.target_position, 3).tolist()} "
+            f"target_leg={np.round(target_joint_positions[leg_slice], 3).tolist()} "
+            f"actual_leg={np.round(actual_joint_positions[leg_slice], 3).tolist()} "
+            f"err_norm={float(np.linalg.norm(joint_error)):.4f} "
+            f"fallen={self.Player.world.global_position[2] < 0.3}"
+        )
+        print(f"waist target={target_joint_positions[10]:.3f}, actual={actual_joint_positions[10]:.3f}")
+
+    def reset(self, seed=None, options=None):
+        '''
+        Reset and stabilize the robot
+        Note: for some behaviors it would be better to reduce stabilization or add noise
+        '''
+        r = self.Player.robot
+        super().reset(seed=seed)
+        if seed is not None:
+            np.random.seed(seed)
+
+        length1 = 2  # randomize target distance
+        length2 = np.random.uniform(0.6, 1)  # randomize target distance
+        length3 = np.random.uniform(0.6, 1)  # randomize target distance
+        angle2 = np.random.uniform(-30, 30)  # randomize initial orientation
+        angle3 = np.random.uniform(-30, 30)  # randomize target direction
+
+        self.step_counter = 0
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Initialize for first step
+        self.walk_cycle_step = 0
+
+        # 随机 beam 目标位置和朝向，增加训练多样性
+        beam_x = (random() - 0.5) * 10
+        beam_y = (random() - 0.5) * 10
+        beam_yaw = uniform(-self.reset_beam_yaw_range_deg, self.reset_beam_yaw_range_deg)
+
+        for _ in range(5):
+            self._safe_receive_world_update(retries=2)
+            self.Player.robot.commit_motor_targets_pd()
+            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=beam_yaw)
+            self.Player.server.send()
+
+        # 执行 Neutral 技能直到完成，给机器人足够时间在 beam 位置稳定站立
+        finished_count = 0
+        for _ in range(50):
+            finished = self.Player.skills_manager.execute("Neutral")
+            self.sync()
+            if finished:
+                finished_count += 1
+                if finished_count >= 20:  # 假设需要连续20次完成才算成功
+                    break
+
+        if self.enable_reset_perturb and self.reset_joint_noise_rad > 0.0:
+            perturb_action = np.zeros(self.no_of_actions, dtype=np.float32)
+            # Perturb waist + lower body only (10:), keep head/arms stable.
+            perturb_action[10:] = np.random.uniform(
+                -self.reset_joint_noise_rad,
+                self.reset_joint_noise_rad,
+                size=(self.no_of_actions - 10,)
+            )
+
+            for _ in range(self.reset_perturb_steps):
+                target_joint_positions = (self.joint_nominal_position + perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+            for i in range(self.reset_recover_steps):
+                # Linearly fade perturbation to help policy start from near-neutral.
+                alpha = 1.0 - float(i + 1) / float(self.reset_recover_steps)
+                target_joint_positions = (self.joint_nominal_position + alpha * perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+        # memory variables
+        self.sync()  
+        self.initial_position = np.array(self.Player.world.global_position[:2])
+        self.previous_pos = self.initial_position.copy()  # Critical: set to actual position
+        self.act = np.zeros(self.no_of_actions, np.float32)
+        # Build target in the robot's current forward direction instead of fixed global +x.
+        heading_deg = float(r.global_orientation_euler[2])
+        forward_offset = MathOps.rotate_2d_vec(np.array([length1, 0.0]), heading_deg, is_rad=False)
+        point1 = self.initial_position + forward_offset
+        point2 = point1 + MathOps.rotate_2d_vec(np.array([length2, 0]), angle2, is_rad=False)
+        point3 = point2 + MathOps.rotate_2d_vec(np.array([length3, 0]), angle3, is_rad=False)
+        self.point_list = [point1]
+        self.target_position = self.point_list[self.waypoint_index]
+        self.initial_height = self.Player.world.global_position[2]
+
+        return self.observe(True), {}
+
+    def render(self, mode='human', close=False):
+        return
+
+    def compute_reward(self, previous_pos, current_pos, action):
+        height = float(self.Player.world.global_position[2])
+
+        orientation_quat_inv = R.from_quat(self.Player.robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+        tilt_mag = float(np.linalg.norm(projected_gravity[:2]))
+        ang_vel = np.deg2rad(self.Player.robot.gyroscope)
+        ang_vel_mag = float(np.linalg.norm(ang_vel))
+
+        is_fallen = height < 0.3
+        if is_fallen:
+            # remain = max(0, 800 - self.step_counter)
+            # return -8.0 - 0.01 * remain
+            return -1.0
+        
+
+
+        # # 目标方向
+        # to_target = self.target_position - current_pos
+        # dist_to_target = float(np.linalg.norm(to_target))
+        # if dist_to_target < 0.5:
+        #     return 15.0
+
+        # forward_dir = to_target / dist_to_target if dist_to_target > 0.1 else np.array([1.0, 0.0])
+        # delta_pos = current_pos - previous_pos
+        # forward_step = float(np.dot(delta_pos, forward_dir))
+        # lateral_step = float(np.linalg.norm(delta_pos - forward_dir * forward_step))
+
+        # 奖励项
+        # progress_reward = 2 * forward_step
+        # lateral_penalty = -0.1 * lateral_step
+        alive_bonus = 2.0
+
+        # action_penalty = -0.01 * float(np.linalg.norm(action))
+        smoothness_penalty = -0.01 * float(np.linalg.norm(action - self.last_action_for_reward))
+
+        posture_penalty = -0.3 * (tilt_mag)
+        ang_vel_penalty = -0.02 * ang_vel_mag
+
+        target_height = self.initial_height
+        height_error =  height - target_height
+        height_penalty = -1 * abs(height_error)  # 惩罚高度偏离，系数可调
+
+        # # 在 compute_reward 开头附近，添加高度变化率计算
+        # if not hasattr(self, 'last_height'):
+        #     self.last_height = height
+        #     self.last_height_time = self.step_counter  # 可选，用于时间间隔
+        # height_rate = height - self.last_height  # 正为上升，负为下降
+        # self.last_height = height
+
+        # 惩罚高度下降（负变化率）
+        # height_down_penalty = -5.0 * max(0, -height_rate)  # 系数可调，-height_rate 为正表示下降幅度
+
+        # # 在 compute_reward 中
+        # if self.step_counter > 50:
+        #     avg_prev_action = np.mean(self.prev_action_history, axis=0)
+        #     novelty = float(np.linalg.norm(action - avg_prev_action))
+        #     exploration_bonus = 0.05 * novelty
+        # else:
+        #     exploration_bonus = 0
+
+        # self.prev_action_history[self.history_idx] = action
+        # self.history_idx = (self.history_idx + 1) % 50
+
+
+        total = (
+            # progress_reward  + 
+                 alive_bonus + 
+                 # lateral_penalty +
+                 # action_penalty +
+                smoothness_penalty +
+                 posture_penalty
+                 + ang_vel_penalty
+                 + height_penalty
+                #  + exploration_bonus
+                # + height_down_penalty
+                 )
+        if time.time() - self.start_time >= 1200:
+            self.start_time = time.time()
+            print(
+                #   f"progress_reward:{progress_reward:.4f}",
+                #   f"lateral_penalty:{lateral_penalty:.4f}",
+                #   f"action_penalty:{action_penalty:.4f}"s, 
+                  f"height_penalty:{height_penalty:.4f}",
+                  f"smoothness_penalty:{smoothness_penalty:.4f},",
+                  f"posture_penalty:{posture_penalty:.4f}",
+                #   f"ang_vel_penalty:{ang_vel_penalty:.4f}",
+                #   f"height_down_penalty:{height_down_penalty:.4f}",
+                #   f"exploration_bonus:{exploration_bonus:.4f}"
+            )
+
+        return total
+
+
+
+    def step(self, action):
+
+        r = self.Player.robot
+        self.previous_action = action
+
+        self.target_joint_positions = (
+                # self.joint_nominal_position +
+                 self.scaling_factor * action
+        )
+        self.target_joint_positions *= self.train_sim_flip
+
+        for idx, target in enumerate(self.target_joint_positions):
+            r.set_motor_target_position(
+                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=40, kd=1.0
+            )
+
+        self.previous_action = action
+
+        self.sync()  # run simulation step
+        self.step_counter += 1
+
+        if self.enable_debug_joint_status and self.step_counter % self.debug_every_n_steps == 0:
+            self.debug_joint_status()
+
+        current_pos = np.array(self.Player.world.global_position[:2], dtype=np.float32)
+
+        # Compute reward based on movement from previous step
+        reward = self.compute_reward(self.previous_pos, current_pos, action)
+
+        # Update previous position
+        self.previous_pos = current_pos.copy()
+        self.last_action_for_reward = action.copy()
+
+        # Fall detection and penalty
+        is_fallen = self.Player.world.global_position[2] < 0.3
+
+        # terminal state: the robot is falling or timeout
+        terminated = is_fallen or self.step_counter > 800 or self.route_completed
+        truncated = False
+
+        return self.observe(), reward, terminated, truncated, {}
+
+
+class Train(Train_Base):
+    def __init__(self, script) -> None:
+        super().__init__(script)
+
+    def train(self, args):
+
+        # --------------------------------------- Learning parameters
+        n_envs = int(os.environ.get("GYM_CPU_N_ENVS", "20"))
+        if n_envs < 1:
+            raise ValueError("GYM_CPU_N_ENVS must be >= 1")
+        server_warmup_sec = float(os.environ.get("GYM_CPU_SERVER_WARMUP_SEC", "3.0"))
+        n_steps_per_env = int(os.environ.get("GYM_CPU_TRAIN_STEPS_PER_ENV", "256"))  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        minibatch_size = int(os.environ.get("GYM_CPU_TRAIN_BATCH_SIZE", "512"))  # should be a factor of (n_steps_per_env * n_envs)
+        total_steps = 30000000
+        learning_rate = float(os.environ.get("GYM_CPU_TRAIN_LR", "3e-4"))
+        folder_name = f'Walk_R{self.robot_type}'
+        model_path = f'./scripts/gyms/logs/{folder_name}/'
+
+        print(f"Model path: {model_path}")
+        print(f"Using {n_envs} parallel environments")
+
+        # --------------------------------------- Run algorithm
+        def init_env(i_env, monitor=False):
+            def thunk():
+                env = WalkEnv(self.ip, self.server_p + i_env)
+                if monitor:
+                    env = Monitor(env)
+                return env
+
+            return thunk
+
+        server_log_dir = os.path.join(model_path, "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1, no_render=True, no_realtime=True)  # include 1 extra server for testing
+
+        # Wait for servers to start
+        print(f"Starting {n_envs + 1} rcssservermj servers...")
+        if server_warmup_sec > 0:
+            print(f"Waiting {server_warmup_sec:.1f}s for server warmup...")
+            sleep(server_warmup_sec)
+        print("Servers started, creating environments...")
+
+        env = SubprocVecEnv([init_env(i, monitor=True) for i in range(n_envs)])
+        # Use single-process eval env to avoid extra subprocess fragility during callback evaluation.
+        eval_env = DummyVecEnv([init_env(n_envs, monitor=True)])
+
+        try:
+            # Custom policy network architecture
+            policy_kwargs = dict(
+                net_arch=dict(
+                    pi=[512, 256, 128],  # Policy network: 3 layers
+                    vf=[512, 256, 128]  # Value network: 3 layers
+                ),
+                activation_fn=__import__('torch.nn', fromlist=['ELU']).ELU,
+            )
+
+            if "model_file" in args:  # retrain
+                model = PPO.load(args["model_file"], env=env, device="cpu", n_envs=n_envs, n_steps=n_steps_per_env,
+                                 batch_size=minibatch_size, learning_rate=learning_rate)
+            else:  # train new model
+                model = PPO(
+                    "MlpPolicy",
+                    env=env,
+                    verbose=1,
+                    n_steps=n_steps_per_env,
+                    batch_size=minibatch_size,
+                    learning_rate=learning_rate,
+                    device="cpu",
+                    policy_kwargs=policy_kwargs,
+                    ent_coef=float(os.environ.get("GYM_CPU_TRAIN_ENT_COEF", "0.05")),  # Entropy coefficient for exploration
+                    clip_range=float(os.environ.get("GYM_CPU_TRAIN_CLIP_RANGE", "0.2")),  # PPO clipping parameter
+                    gae_lambda=0.95,  # GAE lambda
+                    gamma=float(os.environ.get("GYM_CPU_TRAIN_GAMMA", "0.95")), # Discount factor
+                    target_kl=0.03,
+                    n_epochs=int(os.environ.get("GYM_CPU_TRAIN_EPOCHS", "5")),
+                    # tensorboard_log=f"./scripts/gyms/logs/{folder_name}/tensorboard/"
+                )
+
+            model_path = self.learn_model(model, total_steps, model_path, eval_env=eval_env,
+                                          eval_freq=n_steps_per_env * 20, save_freq=n_steps_per_env * 20, eval_eps=100,
+                                          backup_env_file=__file__)
+        except KeyboardInterrupt:
+            sleep(1)  # wait for child processes
+            print("\nctrl+c pressed, aborting...\n")
+            servers.kill()
+            return
+
+        env.close()
+        eval_env.close()
+        servers.kill()
+
+    def test(self, args):
+
+        # Uses different server and monitor ports
+        server_log_dir = os.path.join(args["folder_dir"], "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        test_no_render = os.environ.get("GYM_CPU_TEST_NO_RENDER", "0") == "1"
+        test_no_realtime = os.environ.get("GYM_CPU_TEST_NO_REALTIME", "0") == "1"
+
+        server = Train_Server(
+            self.server_p - 1,
+            self.monitor_p,
+            1,
+            no_render=test_no_render,
+            no_realtime=test_no_realtime,
+        )
+        env = WalkEnv(self.ip, self.server_p - 1)
+        model = PPO.load(args["model_file"], env=env)
+
+        try:
+            self.export_model(args["model_file"], args["model_file"] + ".pkl",
+                              False)  # Export to pkl to create custom behavior
+            self.test_model(model, env, log_path=args["folder_dir"], model_path=args["folder_dir"])
+        except KeyboardInterrupt:
+            print()
+
+        env.close()
+        server.kill()
+
+
+if __name__ == "__main__":
+    from types import SimpleNamespace
+
+    # 创建默认参数
+    script_args = SimpleNamespace(
+        args=SimpleNamespace(
+            i='127.0.0.1',  # Server IP
+            p=3100,  # Server port
+            m=3200,  # Monitor port
+            r=0,  # Robot type
+            t='Gym',  # Team name
+            u=1  # Uniform number
+        )
+    )
+
+    trainer = Train(script_args)
+
+    run_mode = os.environ.get("GYM_CPU_MODE", "train").strip().lower()
+
+    if run_mode == "test":
+        test_model_file = os.environ.get("GYM_CPU_TEST_MODEL", "scripts/gyms/logs/Walk_R0_004/best_model.zip")
+        test_folder = os.environ.get("GYM_CPU_TEST_FOLDER", "scripts/gyms/logs/Walk_R0_004/")
+        trainer.test({"model_file": test_model_file, "folder_dir": test_folder})
+    else:
+        retrain_model = os.environ.get("GYM_CPU_TRAIN_MODEL", "").strip()
+        if retrain_model:
+            trainer.train({"model_file": retrain_model})
+        else:
+            trainer.train({})

scripts/gyms/logs/Walk_R0_008/Walk.py

@@ -0,0 +1,704 @@
+import os
+import numpy as np
+import math
+import time
+from time import sleep
+from random import random
+from random import uniform
+
+from stable_baselines3 import PPO
+from stable_baselines3.common.monitor import Monitor
+from stable_baselines3.common.vec_env import SubprocVecEnv, DummyVecEnv
+
+import gymnasium as gym
+from gymnasium import spaces
+
+from scripts.commons.Train_Base import Train_Base
+from scripts.commons.Server import Server as Train_Server
+
+from agent.base_agent import Base_Agent
+from utils.math_ops import MathOps
+
+from scipy.spatial.transform import Rotation as R
+
+'''
+Objective:
+Learn how to run forward using step primitive
+----------
+- class Basic_Run: implements an OpenAI custom gym
+- class Train:  implements algorithms to train a new model or test an existing model
+'''
+
+
+class WalkEnv(gym.Env):
+    def __init__(self, ip, server_p) -> None:
+
+        # Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
+        self.Player = player = Base_Agent(
+            team_name="Gym",
+            number=1,
+            host=ip,
+            port=server_p
+        )
+        self.robot_type = self.Player.robot
+        self.step_counter = 0  # to limit episode size
+        self.force_play_on = True
+
+        self.target_position = np.array([0.0, 0.0])  # target position in the x-y plane
+        self.initial_position = np.array([0.0, 0.0])  # initial position in the x-y plane
+        self.target_direction = 0.0  # target direction in the x-y plane (relative to the robot's orientation)
+        self.isfallen = False
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.debug_every_n_steps = 5
+        self.enable_debug_joint_status = False
+        self.calibrate_nominal_from_neutral = True
+        self.auto_calibrate_train_sim_flip = True
+        self.nominal_calibrated_once = False
+        self.flip_calibrated_once = False
+        self._target_hz = 0.0
+        self._target_dt = 0.0
+        self._last_sync_time = None
+        target_hz_env = 0
+        if target_hz_env:
+            try:
+                self._target_hz = float(target_hz_env)
+            except ValueError:
+                self._target_hz = 0.0
+        if self._target_hz > 0.0:
+            self._target_dt = 1.0 / self._target_hz
+
+        # State space
+        # 原始观测大小: 78
+        obs_size = 78
+        self.obs = np.zeros(obs_size, np.float32)
+        self.observation_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(obs_size,),
+            dtype=np.float32
+        )
+
+        action_dim = len(self.Player.robot.ROBOT_MOTORS)
+        self.no_of_actions = action_dim
+        self.action_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(action_dim,),
+            dtype=np.float32
+        )
+
+        # 中立姿态
+        self.joint_nominal_position = np.array(
+            [
+                0.0,
+                0.0,
+                0.0,
+                1.4,
+                0.0,
+                -0.4,
+                0.0,
+                -1.4,
+                0.0,
+                0.4,
+                0.0,
+                -0.4,
+                0.0,
+                0.0,
+                0.8,
+                -0.4,
+                0.0,
+                0.4,
+                0.0,
+                0.0,
+                -0.8,
+                0.4,
+                0.0,
+            ]
+        )
+        self.joint_nominal_position = np.zeros(self.no_of_actions)
+        self.train_sim_flip = np.array(
+            [
+                1.0,  # 0: Head_yaw (he1)
+                -1.0,  # 1: Head_pitch (he2)
+                1.0,  # 2: Left_Shoulder_Pitch (lae1)
+                -1.0,  # 3: Left_Shoulder_Roll (lae2)
+                -1.0,  # 4: Left_Elbow_Pitch (lae3)
+                1.0,  # 5: Left_Elbow_Yaw (lae4)
+                -1.0,  # 6: Right_Shoulder_Pitch (rae1)
+                -1.0,  # 7: Right_Shoulder_Roll (rae2)
+                1.0,  # 8: Right_Elbow_Pitch (rae3)
+                1.0,  # 9: Right_Elbow_Yaw (rae4)
+                1.0,  # 10: Waist (te1)
+                1.0,  # 11: Left_Hip_Pitch (lle1)
+                -1.0,  # 12: Left_Hip_Roll (lle2)
+                -1.0,  # 13: Left_Hip_Yaw (lle3)
+                1.0,  # 14: Left_Knee_Pitch (lle4)
+                1.0,  # 15: Left_Ankle_Pitch (lle5)
+                -1.0,  # 16: Left_Ankle_Roll (lle6)
+                -1.0,  # 17: Right_Hip_Pitch (rle1)
+                -1.0,  # 18: Right_Hip_Roll (rle2)
+                -1.0,  # 19: Right_Hip_Yaw (rle3)
+                -1.0,  # 20: Right_Knee_Pitch (rle4)
+                -1.0,  # 21: Right_Ankle_Pitch (rle5)
+                -1.0,  # 22: Right_Ankle_Roll (rle6)
+            ]
+        )
+
+        self.scaling_factor = 0.3
+        # self.scaling_factor = 1
+
+        # Encourage a minimum lateral stance so the policy avoids feet overlap.
+        self.min_stance_rad = 0.10
+
+        # Small reset perturbations for robustness training.
+        self.enable_reset_perturb = False
+        self.reset_beam_yaw_range_deg = 180  # randomize target direction fully to encourage learning a real walk instead of a fixed gait
+        self.reset_joint_noise_rad = 0.015
+        self.reset_perturb_steps = 3
+        self.reset_recover_steps = 8
+
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Track previous position
+        self.Player.server.connect()
+        # sleep(2.0)  # Longer wait for connection to establish completely
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+        self.start_time = time.time()
+
+    def _reconnect_server(self):
+        try:
+            self.Player.server.shutdown()
+        except Exception:
+            pass
+
+        self.Player.server.connect()
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+
+    def _safe_receive_world_update(self, retries=1):
+        last_exc = None
+        for attempt in range(retries + 1):
+            try:
+                self.Player.server.receive()
+                self.Player.world.update()
+                return
+            except (ConnectionResetError, OSError) as exc:
+                last_exc = exc
+                if attempt >= retries:
+                    raise
+                self._reconnect_server()
+        if last_exc is not None:
+            raise last_exc
+
+    def debug_log(self, message):
+        print(message)
+        try:
+            log_path = os.path.join(os.path.dirname(os.path.dirname(__file__)), "comm_debug.log")
+            with open(log_path, "a", encoding="utf-8") as f:
+                f.write(message + "\n")
+        except OSError:
+            pass
+
+    def observe(self, init=False):
+
+        """获取当前观测值"""
+        robot = self.Player.robot
+        world = self.Player.world
+
+        # Safety check: ensure data is available
+
+        # 计算目标速度
+        raw_target = self.target_position - world.global_position[:2]
+        velocity = MathOps.rotate_2d_vec(
+            raw_target,
+            -robot.global_orientation_euler[2],
+            is_rad=False
+        )
+
+        # 计算相对方向
+        rel_orientation = MathOps.vector_angle(velocity) * 0.3
+        rel_orientation = np.clip(rel_orientation, -0.25, 0.25)
+
+        velocity = np.concatenate([velocity, np.array([rel_orientation])])
+        velocity[0] = np.clip(velocity[0], -0.5, 0.5)
+        velocity[1] = np.clip(velocity[1], -0.25, 0.25)
+
+        # 关节状态
+        radian_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        radian_joint_speeds = np.deg2rad(
+            [robot.motor_speeds[motor] for motor in robot.ROBOT_MOTORS]
+        )
+
+        qpos_qvel_previous_action = np.concatenate([
+            (radian_joint_positions * self.train_sim_flip - self.joint_nominal_position) / 4.6,
+            radian_joint_speeds / 110.0 * self.train_sim_flip,
+            self.previous_action / 10.0,
+        ])
+
+        # 角速度
+        ang_vel = np.clip(np.deg2rad(robot.gyroscope) / 50.0, -1.0, 1.0)
+
+        # 投影的重力方向
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+
+        # 组合观测
+        observation = np.concatenate([
+            qpos_qvel_previous_action,
+            ang_vel,
+            velocity,
+            projected_gravity,
+        ])
+
+        observation = np.clip(observation, -10.0, 10.0)
+        return observation.astype(np.float32)
+
+    def sync(self):
+        ''' Run a single simulation step '''
+        self._safe_receive_world_update(retries=1)
+        self.Player.robot.commit_motor_targets_pd()
+        self.Player.server.send()
+        if self._target_dt > 0.0:
+            now = time.time()
+            if self._last_sync_time is None:
+                self._last_sync_time = now
+                return
+            elapsed = now - self._last_sync_time
+            remaining = self._target_dt - elapsed
+            if remaining > 0.0:
+                time.sleep(remaining)
+                now = time.time()
+            self._last_sync_time = now
+
+    def debug_joint_status(self):
+        robot = self.Player.robot
+        actual_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        target_joint_positions = getattr(
+            self,
+            'target_joint_positions',
+            np.zeros(len(robot.ROBOT_MOTORS), dtype=np.float32)
+        )
+        joint_error = actual_joint_positions - target_joint_positions
+        leg_slice = slice(11, None)
+
+        self.debug_log(
+            "[WalkDebug] "
+            f"step={self.step_counter} "
+            f"pos={np.round(self.Player.world.global_position, 3).tolist()} "
+            f"target_xy={np.round(self.target_position, 3).tolist()} "
+            f"target_leg={np.round(target_joint_positions[leg_slice], 3).tolist()} "
+            f"actual_leg={np.round(actual_joint_positions[leg_slice], 3).tolist()} "
+            f"err_norm={float(np.linalg.norm(joint_error)):.4f} "
+            f"fallen={self.Player.world.global_position[2] < 0.3}"
+        )
+        print(f"waist target={target_joint_positions[10]:.3f}, actual={actual_joint_positions[10]:.3f}")
+
+    def reset(self, seed=None, options=None):
+        '''
+        Reset and stabilize the robot
+        Note: for some behaviors it would be better to reduce stabilization or add noise
+        '''
+        r = self.Player.robot
+        super().reset(seed=seed)
+        if seed is not None:
+            np.random.seed(seed)
+
+        length1 = 2  # randomize target distance
+        length2 = np.random.uniform(0.6, 1)  # randomize target distance
+        length3 = np.random.uniform(0.6, 1)  # randomize target distance
+        angle2 = np.random.uniform(-30, 30)  # randomize initial orientation
+        angle3 = np.random.uniform(-30, 30)  # randomize target direction
+
+        self.step_counter = 0
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Initialize for first step
+        self.walk_cycle_step = 0
+
+        # 随机 beam 目标位置和朝向，增加训练多样性
+        beam_x = (random() - 0.5) * 10
+        beam_y = (random() - 0.5) * 10
+        beam_yaw = uniform(-self.reset_beam_yaw_range_deg, self.reset_beam_yaw_range_deg)
+
+        for _ in range(5):
+            self._safe_receive_world_update(retries=2)
+            self.Player.robot.commit_motor_targets_pd()
+            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=beam_yaw)
+            self.Player.server.send()
+
+        # 执行 Neutral 技能直到完成，给机器人足够时间在 beam 位置稳定站立
+        finished_count = 0
+        for _ in range(50):
+            finished = self.Player.skills_manager.execute("Neutral")
+            self.sync()
+            if finished:
+                finished_count += 1
+                if finished_count >= 20:  # 假设需要连续20次完成才算成功
+                    break
+
+        if self.enable_reset_perturb and self.reset_joint_noise_rad > 0.0:
+            perturb_action = np.zeros(self.no_of_actions, dtype=np.float32)
+            # Perturb waist + lower body only (10:), keep head/arms stable.
+            perturb_action[10:] = np.random.uniform(
+                -self.reset_joint_noise_rad,
+                self.reset_joint_noise_rad,
+                size=(self.no_of_actions - 10,)
+            )
+
+            for _ in range(self.reset_perturb_steps):
+                target_joint_positions = (self.joint_nominal_position + perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+            for i in range(self.reset_recover_steps):
+                # Linearly fade perturbation to help policy start from near-neutral.
+                alpha = 1.0 - float(i + 1) / float(self.reset_recover_steps)
+                target_joint_positions = (self.joint_nominal_position + alpha * perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+        # memory variables
+        self.sync()  
+        self.initial_position = np.array(self.Player.world.global_position[:2])
+        self.previous_pos = self.initial_position.copy()  # Critical: set to actual position
+        self.act = np.zeros(self.no_of_actions, np.float32)
+        # Build target in the robot's current forward direction instead of fixed global +x.
+        heading_deg = float(r.global_orientation_euler[2])
+        forward_offset = MathOps.rotate_2d_vec(np.array([length1, 0.0]), heading_deg, is_rad=False)
+        point1 = self.initial_position + forward_offset
+        point2 = point1 + MathOps.rotate_2d_vec(np.array([length2, 0]), angle2, is_rad=False)
+        point3 = point2 + MathOps.rotate_2d_vec(np.array([length3, 0]), angle3, is_rad=False)
+        self.point_list = [point1]
+        self.target_position = self.point_list[self.waypoint_index]
+        self.initial_height = self.Player.world.global_position[2]
+
+        return self.observe(True), {}
+
+    def render(self, mode='human', close=False):
+        return
+
+    def compute_reward(self, previous_pos, current_pos, action):
+        height = float(self.Player.world.global_position[2])
+        robot = self.Player.robot
+
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+        tilt_mag = float(np.linalg.norm(projected_gravity[:2]))
+        ang_vel = np.deg2rad(robot.gyroscope)
+        ang_vel_mag = float(np.linalg.norm(ang_vel))
+
+        is_fallen = height < 0.3
+        if is_fallen:
+            # remain = max(0, 800 - self.step_counter)
+            # return -8.0 - 0.01 * remain
+            return -1.0
+        
+
+
+        # # 目标方向
+        # to_target = self.target_position - current_pos
+        # dist_to_target = float(np.linalg.norm(to_target))
+        # if dist_to_target < 0.5:
+        #     return 15.0
+
+        # forward_dir = to_target / dist_to_target if dist_to_target > 0.1 else np.array([1.0, 0.0])
+        # delta_pos = current_pos - previous_pos
+        # forward_step = float(np.dot(delta_pos, forward_dir))
+        # lateral_step = float(np.linalg.norm(delta_pos - forward_dir * forward_step))
+
+        # 奖励项
+        # progress_reward = 2 * forward_step
+        # lateral_penalty = -0.1 * lateral_step
+        alive_bonus = 2.0
+
+        # action_penalty = -0.01 * float(np.linalg.norm(action))
+        smoothness_penalty = -0.01 * float(np.linalg.norm(action - self.last_action_for_reward))
+
+        posture_penalty = -0.3 * (tilt_mag)
+        ang_vel_penalty = -0.02 * ang_vel_mag
+
+        # Use simulator joint readings in training frame to shape lateral stance.
+        joint_pos = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        ) * self.train_sim_flip
+        left_hip_roll = float(joint_pos[12])
+        right_hip_roll = float(joint_pos[18])
+        left_ankle_roll = float(joint_pos[16])
+        right_ankle_roll = float(joint_pos[22])
+
+        hip_spread = left_hip_roll - right_hip_roll
+        ankle_spread = left_ankle_roll - right_ankle_roll
+        stance_metric = 0.6 * abs(hip_spread) + 0.4 * abs(ankle_spread)
+
+        # Penalize narrow stance (feet too close) and scissoring (cross-leg pattern).
+        stance_collapse_penalty = -4.0 * max(0.0, self.min_stance_rad - stance_metric)
+        cross_leg_penalty = -1.2 * max(0.0, -(hip_spread * ankle_spread))
+
+        target_height = self.initial_height
+        height_error =  height - target_height
+        height_penalty = -0.5 * abs(height_error)  # 惩罚高度偏离，系数可调
+
+        # # 在 compute_reward 开头附近，添加高度变化率计算
+        # if not hasattr(self, 'last_height'):
+        #     self.last_height = height
+        #     self.last_height_time = self.step_counter  # 可选，用于时间间隔
+        # height_rate = height - self.last_height  # 正为上升，负为下降
+        # self.last_height = height
+
+        # 惩罚高度下降（负变化率）
+        # height_down_penalty = -5.0 * max(0, -height_rate)  # 系数可调，-height_rate 为正表示下降幅度
+
+        # # 在 compute_reward 中
+        # if self.step_counter > 50:
+        #     avg_prev_action = np.mean(self.prev_action_history, axis=0)
+        #     novelty = float(np.linalg.norm(action - avg_prev_action))
+        #     exploration_bonus = 0.05 * novelty
+        # else:
+        #     exploration_bonus = 0
+
+        # self.prev_action_history[self.history_idx] = action
+        # self.history_idx = (self.history_idx + 1) % 50
+
+
+        total = (
+            # progress_reward  + 
+                 alive_bonus + 
+                 # lateral_penalty +
+                 # action_penalty +
+                smoothness_penalty +
+                 posture_penalty
+                 + ang_vel_penalty
+                 + height_penalty
+                 + stance_collapse_penalty
+                 + cross_leg_penalty
+                #  + exploration_bonus
+                # + height_down_penalty
+                 )
+        if time.time() - self.start_time >= 600:
+            self.start_time = time.time()
+            print(
+                #   f"progress_reward:{progress_reward:.4f}",
+                #   f"lateral_penalty:{lateral_penalty:.4f}",
+                #   f"action_penalty:{action_penalty:.4f}"s, 
+                  f"height_penalty:{height_penalty:.4f}",
+                  f"smoothness_penalty:{smoothness_penalty:.4f},",
+                  f"posture_penalty:{posture_penalty:.4f}",
+                  f"stance_collapse_penalty:{stance_collapse_penalty:.4f}",
+                  f"cross_leg_penalty:{cross_leg_penalty:.4f}",
+                #   f"ang_vel_penalty:{ang_vel_penalty:.4f}",
+                #   f"height_down_penalty:{height_down_penalty:.4f}",
+                #   f"exploration_bonus:{exploration_bonus:.4f}"
+            )
+
+        return total
+
+
+
+    def step(self, action):
+
+        r = self.Player.robot
+        self.previous_action = action
+
+        self.target_joint_positions = (
+                # self.joint_nominal_position +
+                 self.scaling_factor * action
+        )
+        self.target_joint_positions *= self.train_sim_flip
+
+        for idx, target in enumerate(self.target_joint_positions):
+            r.set_motor_target_position(
+                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=40, kd=1.0
+            )
+
+        self.previous_action = action
+
+        self.sync()  # run simulation step
+        self.step_counter += 1
+
+        if self.enable_debug_joint_status and self.step_counter % self.debug_every_n_steps == 0:
+            self.debug_joint_status()
+
+        current_pos = np.array(self.Player.world.global_position[:2], dtype=np.float32)
+
+        # Compute reward based on movement from previous step
+        reward = self.compute_reward(self.previous_pos, current_pos, action)
+
+        # Update previous position
+        self.previous_pos = current_pos.copy()
+        self.last_action_for_reward = action.copy()
+
+        # Fall detection and penalty
+        is_fallen = self.Player.world.global_position[2] < 0.3
+
+        # terminal state: the robot is falling or timeout
+        terminated = is_fallen or self.step_counter > 800 or self.route_completed
+        truncated = False
+
+        return self.observe(), reward, terminated, truncated, {}
+
+
+class Train(Train_Base):
+    def __init__(self, script) -> None:
+        super().__init__(script)
+
+    def train(self, args):
+
+        # --------------------------------------- Learning parameters
+        n_envs = int(os.environ.get("GYM_CPU_N_ENVS", "20"))
+        if n_envs < 1:
+            raise ValueError("GYM_CPU_N_ENVS must be >= 1")
+        server_warmup_sec = float(os.environ.get("GYM_CPU_SERVER_WARMUP_SEC", "3.0"))
+        n_steps_per_env = int(os.environ.get("GYM_CPU_TRAIN_STEPS_PER_ENV", "256"))  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        minibatch_size = int(os.environ.get("GYM_CPU_TRAIN_BATCH_SIZE", "512"))  # should be a factor of (n_steps_per_env * n_envs)
+        total_steps = 30000000
+        learning_rate = float(os.environ.get("GYM_CPU_TRAIN_LR", "3e-4"))
+        folder_name = f'Walk_R{self.robot_type}'
+        model_path = f'./scripts/gyms/logs/{folder_name}/'
+
+        print(f"Model path: {model_path}")
+        print(f"Using {n_envs} parallel environments")
+
+        # --------------------------------------- Run algorithm
+        def init_env(i_env, monitor=False):
+            def thunk():
+                env = WalkEnv(self.ip, self.server_p + i_env)
+                if monitor:
+                    env = Monitor(env)
+                return env
+
+            return thunk
+
+        server_log_dir = os.path.join(model_path, "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1, no_render=True, no_realtime=True)  # include 1 extra server for testing
+
+        # Wait for servers to start
+        print(f"Starting {n_envs + 1} rcssservermj servers...")
+        if server_warmup_sec > 0:
+            print(f"Waiting {server_warmup_sec:.1f}s for server warmup...")
+            sleep(server_warmup_sec)
+        print("Servers started, creating environments...")
+
+        env = SubprocVecEnv([init_env(i, monitor=True) for i in range(n_envs)])
+        # Use single-process eval env to avoid extra subprocess fragility during callback evaluation.
+        eval_env = DummyVecEnv([init_env(n_envs, monitor=True)])
+
+        try:
+            # Custom policy network architecture
+            policy_kwargs = dict(
+                net_arch=dict(
+                    pi=[512, 256, 128],  # Policy network: 3 layers
+                    vf=[512, 256, 128]  # Value network: 3 layers
+                ),
+                activation_fn=__import__('torch.nn', fromlist=['ELU']).ELU,
+            )
+
+            if "model_file" in args:  # retrain
+                model = PPO.load(args["model_file"], env=env, device="cpu", n_envs=n_envs, n_steps=n_steps_per_env,
+                                 batch_size=minibatch_size, learning_rate=learning_rate)
+            else:  # train new model
+                model = PPO(
+                    "MlpPolicy",
+                    env=env,
+                    verbose=1,
+                    n_steps=n_steps_per_env,
+                    batch_size=minibatch_size,
+                    learning_rate=learning_rate,
+                    device="cpu",
+                    policy_kwargs=policy_kwargs,
+                    ent_coef=float(os.environ.get("GYM_CPU_TRAIN_ENT_COEF", "0.05")),  # Entropy coefficient for exploration
+                    clip_range=float(os.environ.get("GYM_CPU_TRAIN_CLIP_RANGE", "0.2")),  # PPO clipping parameter
+                    gae_lambda=0.95,  # GAE lambda
+                    gamma=float(os.environ.get("GYM_CPU_TRAIN_GAMMA", "0.95")), # Discount factor
+                    target_kl=0.03,
+                    n_epochs=int(os.environ.get("GYM_CPU_TRAIN_EPOCHS", "5")),
+                    tensorboard_log=f"./scripts/gyms/logs/{folder_name}/tensorboard/"
+                )
+
+            model_path = self.learn_model(model, total_steps, model_path, eval_env=eval_env,
+                                          eval_freq=n_steps_per_env * 20, save_freq=n_steps_per_env * 20, eval_eps=100,
+                                          backup_env_file=__file__)
+        except KeyboardInterrupt:
+            sleep(1)  # wait for child processes
+            print("\nctrl+c pressed, aborting...\n")
+            servers.kill()
+            return
+
+        env.close()
+        eval_env.close()
+        servers.kill()
+
+    def test(self, args):
+
+        # Uses different server and monitor ports
+        server_log_dir = os.path.join(args["folder_dir"], "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        test_no_render = os.environ.get("GYM_CPU_TEST_NO_RENDER", "0") == "1"
+        test_no_realtime = os.environ.get("GYM_CPU_TEST_NO_REALTIME", "0") == "1"
+
+        server = Train_Server(
+            self.server_p - 1,
+            self.monitor_p,
+            1,
+            no_render=test_no_render,
+            no_realtime=test_no_realtime,
+        )
+        env = WalkEnv(self.ip, self.server_p - 1)
+        model = PPO.load(args["model_file"], env=env)
+
+        try:
+            self.export_model(args["model_file"], args["model_file"] + ".pkl",
+                              False)  # Export to pkl to create custom behavior
+            self.test_model(model, env, log_path=args["folder_dir"], model_path=args["folder_dir"])
+        except KeyboardInterrupt:
+            print()
+
+        env.close()
+        server.kill()
+
+
+if __name__ == "__main__":
+    from types import SimpleNamespace
+
+    # 创建默认参数
+    script_args = SimpleNamespace(
+        args=SimpleNamespace(
+            i='127.0.0.1',  # Server IP
+            p=3100,  # Server port
+            m=3200,  # Monitor port
+            r=0,  # Robot type
+            t='Gym',  # Team name
+            u=1  # Uniform number
+        )
+    )
+
+    trainer = Train(script_args)
+
+    run_mode = os.environ.get("GYM_CPU_MODE", "train").strip().lower()
+
+    if run_mode == "test":
+        test_model_file = os.environ.get("GYM_CPU_TEST_MODEL", "scripts/gyms/logs/Walk_R0_004/best_model.zip")
+        test_folder = os.environ.get("GYM_CPU_TEST_FOLDER", "scripts/gyms/logs/Walk_R0_004/")
+        trainer.test({"model_file": test_model_file, "folder_dir": test_folder})
+    else:
+        retrain_model = os.environ.get("GYM_CPU_TRAIN_MODEL", "").strip()
+        if retrain_model:
+            trainer.train({"model_file": retrain_model})
+        else:
+            trainer.train({})

scripts/gyms/logs/Walk_R0_009/Walk.py

@@ -0,0 +1,705 @@
+import os
+import numpy as np
+import math
+import time
+from time import sleep
+from random import random
+from random import uniform
+from itertools import count
+
+from stable_baselines3 import PPO
+from stable_baselines3.common.monitor import Monitor
+from stable_baselines3.common.vec_env import SubprocVecEnv, DummyVecEnv
+
+import gymnasium as gym
+from gymnasium import spaces
+
+from scripts.commons.Train_Base import Train_Base
+from scripts.commons.Server import Server as Train_Server
+
+from agent.base_agent import Base_Agent
+from utils.math_ops import MathOps
+
+from scipy.spatial.transform import Rotation as R
+
+'''
+Objective:
+Learn how to run forward using step primitive
+----------
+- class Basic_Run: implements an OpenAI custom gym
+- class Train:  implements algorithms to train a new model or test an existing model
+'''
+
+
+class WalkEnv(gym.Env):
+    def __init__(self, ip, server_p) -> None:
+
+        # Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
+        self.Player = player = Base_Agent(
+            team_name="Gym",
+            number=1,
+            host=ip,
+            port=server_p
+        )
+        self.robot_type = self.Player.robot
+        self.step_counter = 0  # to limit episode size
+        self.force_play_on = True
+
+        self.target_position = np.array([0.0, 0.0])  # target position in the x-y plane
+        self.initial_position = np.array([0.0, 0.0])  # initial position in the x-y plane
+        self.target_direction = 0.0  # target direction in the x-y plane (relative to the robot's orientation)
+        self.isfallen = False
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.debug_every_n_steps = 5
+        self.enable_debug_joint_status = False
+        self.calibrate_nominal_from_neutral = True
+        self.auto_calibrate_train_sim_flip = True
+        self.nominal_calibrated_once = False
+        self.flip_calibrated_once = False
+        self._target_hz = 0.0
+        self._target_dt = 0.0
+        self._last_sync_time = None
+        target_hz_env = 0
+        if target_hz_env:
+            try:
+                self._target_hz = float(target_hz_env)
+            except ValueError:
+                self._target_hz = 0.0
+        if self._target_hz > 0.0:
+            self._target_dt = 1.0 / self._target_hz
+
+        # State space
+        # 原始观测大小: 78
+        obs_size = 78
+        self.obs = np.zeros(obs_size, np.float32)
+        self.observation_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(obs_size,),
+            dtype=np.float32
+        )
+
+        action_dim = len(self.Player.robot.ROBOT_MOTORS)
+        self.no_of_actions = action_dim
+        self.action_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(action_dim,),
+            dtype=np.float32
+        )
+
+        # 中立姿态
+        self.joint_nominal_position = np.array(
+            [
+                0.0,
+                0.0,
+                0.0,
+                1.4,
+                0.0,
+                -0.4,
+                0.0,
+                -1.4,
+                0.0,
+                0.4,
+                0.0,
+                -0.4,
+                0.0,
+                0.0,
+                0.8,
+                -0.4,
+                0.0,
+                0.4,
+                0.0,
+                0.0,
+                -0.8,
+                0.4,
+                0.0,
+            ]
+        )
+        self.joint_nominal_position = np.zeros(self.no_of_actions)
+        self.train_sim_flip = np.array(
+            [
+                1.0,  # 0: Head_yaw (he1)
+                -1.0,  # 1: Head_pitch (he2)
+                1.0,  # 2: Left_Shoulder_Pitch (lae1)
+                -1.0,  # 3: Left_Shoulder_Roll (lae2)
+                -1.0,  # 4: Left_Elbow_Pitch (lae3)
+                1.0,  # 5: Left_Elbow_Yaw (lae4)
+                -1.0,  # 6: Right_Shoulder_Pitch (rae1)
+                -1.0,  # 7: Right_Shoulder_Roll (rae2)
+                1.0,  # 8: Right_Elbow_Pitch (rae3)
+                1.0,  # 9: Right_Elbow_Yaw (rae4)
+                1.0,  # 10: Waist (te1)
+                1.0,  # 11: Left_Hip_Pitch (lle1)
+                -1.0,  # 12: Left_Hip_Roll (lle2)
+                -1.0,  # 13: Left_Hip_Yaw (lle3)
+                1.0,  # 14: Left_Knee_Pitch (lle4)
+                1.0,  # 15: Left_Ankle_Pitch (lle5)
+                -1.0,  # 16: Left_Ankle_Roll (lle6)
+                -1.0,  # 17: Right_Hip_Pitch (rle1)
+                -1.0,  # 18: Right_Hip_Roll (rle2)
+                -1.0,  # 19: Right_Hip_Yaw (rle3)
+                -1.0,  # 20: Right_Knee_Pitch (rle4)
+                -1.0,  # 21: Right_Ankle_Pitch (rle5)
+                -1.0,  # 22: Right_Ankle_Roll (rle6)
+            ]
+        )
+
+        self.scaling_factor = 0.3
+        # self.scaling_factor = 1
+
+        # Encourage a minimum lateral stance so the policy avoids feet overlap.
+        self.min_stance_rad = 0.10
+
+        # Small reset perturbations for robustness training.
+        self.enable_reset_perturb = False
+        self.reset_beam_yaw_range_deg = 180  # randomize target direction fully to encourage learning a real walk instead of a fixed gait
+        self.reset_joint_noise_rad = 0.015
+        self.reset_perturb_steps = 3
+        self.reset_recover_steps = 8
+
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Track previous position
+        self.Player.server.connect()
+        # sleep(2.0)  # Longer wait for connection to establish completely
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+        self.start_time = time.time()
+
+    def _reconnect_server(self):
+        try:
+            self.Player.server.shutdown()
+        except Exception:
+            pass
+
+        self.Player.server.connect()
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+
+    def _safe_receive_world_update(self, retries=1):
+        last_exc = None
+        for attempt in range(retries + 1):
+            try:
+                self.Player.server.receive()
+                self.Player.world.update()
+                return
+            except (ConnectionResetError, OSError) as exc:
+                last_exc = exc
+                if attempt >= retries:
+                    raise
+                self._reconnect_server()
+        if last_exc is not None:
+            raise last_exc
+
+    def debug_log(self, message):
+        print(message)
+        try:
+            log_path = os.path.join(os.path.dirname(os.path.dirname(__file__)), "comm_debug.log")
+            with open(log_path, "a", encoding="utf-8") as f:
+                f.write(message + "\n")
+        except OSError:
+            pass
+
+    def observe(self, init=False):
+
+        """获取当前观测值"""
+        robot = self.Player.robot
+        world = self.Player.world
+
+        # Safety check: ensure data is available
+
+        # 计算目标速度
+        raw_target = self.target_position - world.global_position[:2]
+        velocity = MathOps.rotate_2d_vec(
+            raw_target,
+            -robot.global_orientation_euler[2],
+            is_rad=False
+        )
+
+        # 计算相对方向
+        rel_orientation = MathOps.vector_angle(velocity) * 0.3
+        rel_orientation = np.clip(rel_orientation, -0.25, 0.25)
+
+        velocity = np.concatenate([velocity, np.array([rel_orientation])])
+        velocity[0] = np.clip(velocity[0], -0.5, 0.5)
+        velocity[1] = np.clip(velocity[1], -0.25, 0.25)
+
+        # 关节状态
+        radian_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        radian_joint_speeds = np.deg2rad(
+            [robot.motor_speeds[motor] for motor in robot.ROBOT_MOTORS]
+        )
+
+        qpos_qvel_previous_action = np.concatenate([
+            (radian_joint_positions * self.train_sim_flip - self.joint_nominal_position) / 4.6,
+            radian_joint_speeds / 110.0 * self.train_sim_flip,
+            self.previous_action / 10.0,
+        ])
+
+        # 角速度
+        ang_vel = np.clip(np.deg2rad(robot.gyroscope) / 50.0, -1.0, 1.0)
+
+        # 投影的重力方向
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+
+        # 组合观测
+        observation = np.concatenate([
+            qpos_qvel_previous_action,
+            ang_vel,
+            velocity,
+            projected_gravity,
+        ])
+
+        observation = np.clip(observation, -10.0, 10.0)
+        return observation.astype(np.float32)
+
+    def sync(self):
+        ''' Run a single simulation step '''
+        self._safe_receive_world_update(retries=1)
+        self.Player.robot.commit_motor_targets_pd()
+        self.Player.server.send()
+        if self._target_dt > 0.0:
+            now = time.time()
+            if self._last_sync_time is None:
+                self._last_sync_time = now
+                return
+            elapsed = now - self._last_sync_time
+            remaining = self._target_dt - elapsed
+            if remaining > 0.0:
+                time.sleep(remaining)
+                now = time.time()
+            self._last_sync_time = now
+
+    def debug_joint_status(self):
+        robot = self.Player.robot
+        actual_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        target_joint_positions = getattr(
+            self,
+            'target_joint_positions',
+            np.zeros(len(robot.ROBOT_MOTORS), dtype=np.float32)
+        )
+        joint_error = actual_joint_positions - target_joint_positions
+        leg_slice = slice(11, None)
+
+        self.debug_log(
+            "[WalkDebug] "
+            f"step={self.step_counter} "
+            f"pos={np.round(self.Player.world.global_position, 3).tolist()} "
+            f"target_xy={np.round(self.target_position, 3).tolist()} "
+            f"target_leg={np.round(target_joint_positions[leg_slice], 3).tolist()} "
+            f"actual_leg={np.round(actual_joint_positions[leg_slice], 3).tolist()} "
+            f"err_norm={float(np.linalg.norm(joint_error)):.4f} "
+            f"fallen={self.Player.world.global_position[2] < 0.3}"
+        )
+        print(f"waist target={target_joint_positions[10]:.3f}, actual={actual_joint_positions[10]:.3f}")
+
+    def reset(self, seed=None, options=None):
+        '''
+        Reset and stabilize the robot
+        Note: for some behaviors it would be better to reduce stabilization or add noise
+        '''
+        r = self.Player.robot
+        super().reset(seed=seed)
+        if seed is not None:
+            np.random.seed(seed)
+
+        length1 = 2  # randomize target distance
+        length2 = np.random.uniform(0.6, 1)  # randomize target distance
+        length3 = np.random.uniform(0.6, 1)  # randomize target distance
+        angle2 = np.random.uniform(-30, 30)  # randomize initial orientation
+        angle3 = np.random.uniform(-30, 30)  # randomize target direction
+
+        self.step_counter = 0
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Initialize for first step
+        self.walk_cycle_step = 0
+
+        # 随机 beam 目标位置和朝向，增加训练多样性
+        beam_x = (random() - 0.5) * 10
+        beam_y = (random() - 0.5) * 10
+        beam_yaw = uniform(-self.reset_beam_yaw_range_deg, self.reset_beam_yaw_range_deg)
+
+        for _ in range(5):
+            self._safe_receive_world_update(retries=2)
+            self.Player.robot.commit_motor_targets_pd()
+            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=beam_yaw)
+            self.Player.server.send()
+
+        # 执行 Neutral 技能直到完成，给机器人足够时间在 beam 位置稳定站立
+        finished_count = 0
+        for _ in range(50):
+            finished = self.Player.skills_manager.execute("Neutral")
+            self.sync()
+            if finished:
+                finished_count += 1
+                if finished_count >= 20:  # 假设需要连续20次完成才算成功
+                    break
+
+        if self.enable_reset_perturb and self.reset_joint_noise_rad > 0.0:
+            perturb_action = np.zeros(self.no_of_actions, dtype=np.float32)
+            # Perturb waist + lower body only (10:), keep head/arms stable.
+            perturb_action[10:] = np.random.uniform(
+                -self.reset_joint_noise_rad,
+                self.reset_joint_noise_rad,
+                size=(self.no_of_actions - 10,)
+            )
+
+            for _ in range(self.reset_perturb_steps):
+                target_joint_positions = (self.joint_nominal_position + perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+            for i in range(self.reset_recover_steps):
+                # Linearly fade perturbation to help policy start from near-neutral.
+                alpha = 1.0 - float(i + 1) / float(self.reset_recover_steps)
+                target_joint_positions = (self.joint_nominal_position + alpha * perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+        # memory variables
+        self.sync()  
+        self.initial_position = np.array(self.Player.world.global_position[:2])
+        self.previous_pos = self.initial_position.copy()  # Critical: set to actual position
+        self.act = np.zeros(self.no_of_actions, np.float32)
+        # Build target in the robot's current forward direction instead of fixed global +x.
+        heading_deg = float(r.global_orientation_euler[2])
+        forward_offset = MathOps.rotate_2d_vec(np.array([length1, 0.0]), heading_deg, is_rad=False)
+        point1 = self.initial_position + forward_offset
+        point2 = point1 + MathOps.rotate_2d_vec(np.array([length2, 0]), angle2, is_rad=False)
+        point3 = point2 + MathOps.rotate_2d_vec(np.array([length3, 0]), angle3, is_rad=False)
+        self.point_list = [point1]
+        self.target_position = self.point_list[self.waypoint_index]
+        self.initial_height = self.Player.world.global_position[2]
+
+        return self.observe(True), {}
+
+    def render(self, mode='human', close=False):
+        return
+
+    def compute_reward(self, previous_pos, current_pos, action):
+        height = float(self.Player.world.global_position[2])
+        robot = self.Player.robot
+
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+        tilt_mag = float(np.linalg.norm(projected_gravity[:2]))
+        ang_vel = np.deg2rad(robot.gyroscope)
+        ang_vel_mag = float(np.linalg.norm(ang_vel))
+
+        is_fallen = height < 0.3
+        if is_fallen:
+            # remain = max(0, 800 - self.step_counter)
+            # return -8.0 - 0.01 * remain
+            return -1.0
+        
+
+
+        # # 目标方向
+        # to_target = self.target_position - current_pos
+        # dist_to_target = float(np.linalg.norm(to_target))
+        # if dist_to_target < 0.5:
+        #     return 15.0
+
+        # forward_dir = to_target / dist_to_target if dist_to_target > 0.1 else np.array([1.0, 0.0])
+        # delta_pos = current_pos - previous_pos
+        # forward_step = float(np.dot(delta_pos, forward_dir))
+        # lateral_step = float(np.linalg.norm(delta_pos - forward_dir * forward_step))
+
+        # 奖励项
+        # progress_reward = 2 * forward_step
+        # lateral_penalty = -0.1 * lateral_step
+        alive_bonus = 2.0
+
+        # action_penalty = -0.01 * float(np.linalg.norm(action))
+        smoothness_penalty = -0.01 * float(np.linalg.norm(action - self.last_action_for_reward))
+
+        posture_penalty = -0.3 * (tilt_mag)
+        ang_vel_penalty = -0.02 * ang_vel_mag
+
+        # Use simulator joint readings in training frame to shape lateral stance.
+        joint_pos = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        ) * self.train_sim_flip
+        left_hip_roll = float(joint_pos[12])
+        right_hip_roll = float(joint_pos[18])
+        left_ankle_roll = float(joint_pos[16])
+        right_ankle_roll = float(joint_pos[22])
+
+        hip_spread = left_hip_roll - right_hip_roll
+        ankle_spread = left_ankle_roll - right_ankle_roll
+        stance_metric = 0.6 * abs(hip_spread) + 0.4 * abs(ankle_spread)
+
+        # Penalize narrow stance (feet too close) and scissoring (cross-leg pattern).
+        stance_collapse_penalty = -4.0 * max(0.0, self.min_stance_rad - stance_metric)
+        cross_leg_penalty = -1.2 * max(0.0, -(hip_spread * ankle_spread))
+
+        target_height = self.initial_height
+        height_error =  height - target_height
+        height_penalty = -0.5 * abs(height_error)  # 惩罚高度偏离，系数可调
+
+        # # 在 compute_reward 开头附近，添加高度变化率计算
+        # if not hasattr(self, 'last_height'):
+        #     self.last_height = height
+        #     self.last_height_time = self.step_counter  # 可选，用于时间间隔
+        # height_rate = height - self.last_height  # 正为上升，负为下降
+        # self.last_height = height
+
+        # 惩罚高度下降（负变化率）
+        # height_down_penalty = -5.0 * max(0, -height_rate)  # 系数可调，-height_rate 为正表示下降幅度
+
+        # # 在 compute_reward 中
+        # if self.step_counter > 50:
+        #     avg_prev_action = np.mean(self.prev_action_history, axis=0)
+        #     novelty = float(np.linalg.norm(action - avg_prev_action))
+        #     exploration_bonus = 0.05 * novelty
+        # else:
+        #     exploration_bonus = 0
+
+        # self.prev_action_history[self.history_idx] = action
+        # self.history_idx = (self.history_idx + 1) % 50
+
+
+        total = (
+            # progress_reward  + 
+                 alive_bonus + 
+                 # lateral_penalty +
+                 # action_penalty +
+                smoothness_penalty +
+                 posture_penalty
+                 + ang_vel_penalty
+                 + height_penalty
+                 + stance_collapse_penalty
+                 + cross_leg_penalty
+                #  + exploration_bonus
+                # + height_down_penalty
+                 )
+        if time.time() - self.start_time >= 600:
+            self.start_time = time.time()
+            print(
+                #   f"progress_reward:{progress_reward:.4f}",
+                #   f"lateral_penalty:{lateral_penalty:.4f}",
+                #   f"action_penalty:{action_penalty:.4f}"s, 
+                  f"height_penalty:{height_penalty:.4f}",
+                  f"smoothness_penalty:{smoothness_penalty:.4f},",
+                  f"posture_penalty:{posture_penalty:.4f}",
+                  f"stance_collapse_penalty:{stance_collapse_penalty:.4f}",
+                  f"cross_leg_penalty:{cross_leg_penalty:.4f}",
+                #   f"ang_vel_penalty:{ang_vel_penalty:.4f}",
+                #   f"height_down_penalty:{height_down_penalty:.4f}",
+                #   f"exploration_bonus:{exploration_bonus:.4f}"
+            )
+
+        return total
+
+
+
+    def step(self, action):
+
+        r = self.Player.robot
+        self.previous_action = action
+
+        self.target_joint_positions = (
+                # self.joint_nominal_position +
+                 self.scaling_factor * action
+        )
+        self.target_joint_positions *= self.train_sim_flip
+
+        for idx, target in enumerate(self.target_joint_positions):
+            r.set_motor_target_position(
+                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=40, kd=1.0
+            )
+
+        self.previous_action = action
+
+        self.sync()  # run simulation step
+        self.step_counter += 1
+
+        if self.enable_debug_joint_status and self.step_counter % self.debug_every_n_steps == 0:
+            self.debug_joint_status()
+
+        current_pos = np.array(self.Player.world.global_position[:2], dtype=np.float32)
+
+        # Compute reward based on movement from previous step
+        reward = self.compute_reward(self.previous_pos, current_pos, action)
+
+        # Update previous position
+        self.previous_pos = current_pos.copy()
+        self.last_action_for_reward = action.copy()
+
+        # Fall detection and penalty
+        is_fallen = self.Player.world.global_position[2] < 0.3
+
+        # terminal state: the robot is falling or timeout
+        terminated = is_fallen or self.step_counter > 800 or self.route_completed
+        truncated = False
+
+        return self.observe(), reward, terminated, truncated, {}
+
+
+class Train(Train_Base):
+    def __init__(self, script) -> None:
+        super().__init__(script)
+
+    def train(self, args):
+
+        # --------------------------------------- Learning parameters
+        n_envs = int(os.environ.get("GYM_CPU_N_ENVS", "20"))
+        if n_envs < 1:
+            raise ValueError("GYM_CPU_N_ENVS must be >= 1")
+        server_warmup_sec = float(os.environ.get("GYM_CPU_SERVER_WARMUP_SEC", "3.0"))
+        n_steps_per_env = int(os.environ.get("GYM_CPU_TRAIN_STEPS_PER_ENV", "256"))  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        minibatch_size = int(os.environ.get("GYM_CPU_TRAIN_BATCH_SIZE", "512"))  # should be a factor of (n_steps_per_env * n_envs)
+        total_steps = 30000000
+        learning_rate = float(os.environ.get("GYM_CPU_TRAIN_LR", "3e-4"))
+        folder_name = f'Walk_R{self.robot_type}'
+        model_path = f'./scripts/gyms/logs/{folder_name}/'
+
+        print(f"Model path: {model_path}")
+        print(f"Using {n_envs} parallel environments")
+
+        # --------------------------------------- Run algorithm
+        def init_env(i_env, monitor=False):
+            def thunk():
+                env = WalkEnv(self.ip, self.server_p + i_env)
+                if monitor:
+                    env = Monitor(env)
+                return env
+
+            return thunk
+
+        server_log_dir = os.path.join(model_path, "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1, no_render=True, no_realtime=True)  # include 1 extra server for testing
+
+        # Wait for servers to start
+        print(f"Starting {n_envs + 1} rcssservermj servers...")
+        if server_warmup_sec > 0:
+            print(f"Waiting {server_warmup_sec:.1f}s for server warmup...")
+            sleep(server_warmup_sec)
+        print("Servers started, creating environments...")
+
+        env = SubprocVecEnv([init_env(i, monitor=True) for i in range(n_envs)])
+        # Use single-process eval env to avoid extra subprocess fragility during callback evaluation.
+        eval_env = DummyVecEnv([init_env(n_envs, monitor=True)])
+
+        try:
+            # Custom policy network architecture
+            policy_kwargs = dict(
+                net_arch=dict(
+                    pi=[512, 256, 128],  # Policy network: 3 layers
+                    vf=[512, 256, 128]  # Value network: 3 layers
+                ),
+                activation_fn=__import__('torch.nn', fromlist=['ELU']).ELU,
+            )
+
+            if "model_file" in args:  # retrain
+                model = PPO.load(args["model_file"], env=env, device="cpu", n_envs=n_envs, n_steps=n_steps_per_env,
+                                 batch_size=minibatch_size, learning_rate=learning_rate)
+            else:  # train new model
+                model = PPO(
+                    "MlpPolicy",
+                    env=env,
+                    verbose=1,
+                    n_steps=n_steps_per_env,
+                    batch_size=minibatch_size,
+                    learning_rate=learning_rate,
+                    device="cpu",
+                    policy_kwargs=policy_kwargs,
+                    ent_coef=float(os.environ.get("GYM_CPU_TRAIN_ENT_COEF", "0.05")),  # Entropy coefficient for exploration
+                    clip_range=float(os.environ.get("GYM_CPU_TRAIN_CLIP_RANGE", "0.2")),  # PPO clipping parameter
+                    gae_lambda=0.95,  # GAE lambda
+                    gamma=float(os.environ.get("GYM_CPU_TRAIN_GAMMA", "0.95")), # Discount factor
+                    # target_kl=0.03,
+                    n_epochs=int(os.environ.get("GYM_CPU_TRAIN_EPOCHS", "5")),
+                    tensorboard_log=f"./scripts/gyms/logs/{folder_name}/tensorboard/"
+                )
+
+            model_path = self.learn_model(model, total_steps, model_path, eval_env=eval_env,
+                                          eval_freq=n_steps_per_env * 20, save_freq=n_steps_per_env * 20, eval_eps=100,
+                                          backup_env_file=__file__)
+        except KeyboardInterrupt:
+            sleep(1)  # wait for child processes
+            print("\nctrl+c pressed, aborting...\n")
+            servers.kill()
+            return
+
+        env.close()
+        eval_env.close()
+        servers.kill()
+
+    def test(self, args):
+
+        # Uses different server and monitor ports
+        server_log_dir = os.path.join(args["folder_dir"], "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        test_no_render = os.environ.get("GYM_CPU_TEST_NO_RENDER", "0") == "1"
+        test_no_realtime = os.environ.get("GYM_CPU_TEST_NO_REALTIME", "0") == "1"
+
+        server = Train_Server(
+            self.server_p - 1,
+            self.monitor_p,
+            1,
+            no_render=test_no_render,
+            no_realtime=test_no_realtime,
+        )
+        env = WalkEnv(self.ip, self.server_p - 1)
+        model = PPO.load(args["model_file"], env=env)
+
+        try:
+            self.export_model(args["model_file"], args["model_file"] + ".pkl",
+                              False)  # Export to pkl to create custom behavior
+            self.test_model(model, env, log_path=args["folder_dir"], model_path=args["folder_dir"])
+        except KeyboardInterrupt:
+            print()
+
+        env.close()
+        server.kill()
+
+
+if __name__ == "__main__":
+    from types import SimpleNamespace
+
+    # 创建默认参数
+    script_args = SimpleNamespace(
+        args=SimpleNamespace(
+            i='127.0.0.1',  # Server IP
+            p=3100,  # Server port
+            m=3200,  # Monitor port
+            r=0,  # Robot type
+            t='Gym',  # Team name
+            u=1  # Uniform number
+        )
+    )
+
+    trainer = Train(script_args)
+
+    run_mode = os.environ.get("GYM_CPU_MODE", "train").strip().lower()
+
+    if run_mode == "test":
+        test_model_file = os.environ.get("GYM_CPU_TEST_MODEL", "scripts/gyms/logs/Walk_R0_004/best_model.zip")
+        test_folder = os.environ.get("GYM_CPU_TEST_FOLDER", "scripts/gyms/logs/Walk_R0_004/")
+        trainer.test({"model_file": test_model_file, "folder_dir": test_folder})
+    else:
+        retrain_model = os.environ.get("GYM_CPU_TRAIN_MODEL", "").strip()
+        if retrain_model:
+            trainer.train({"model_file": retrain_model})
+        else:
+            trainer.train({})

scripts/gyms/logs/Walk_R0_010/Walk.py

@@ -0,0 +1,705 @@
+import os
+import numpy as np
+import math
+import time
+from time import sleep
+from random import random
+from random import uniform
+from itertools import count
+
+from stable_baselines3 import PPO
+from stable_baselines3.common.monitor import Monitor
+from stable_baselines3.common.vec_env import SubprocVecEnv, DummyVecEnv
+
+import gymnasium as gym
+from gymnasium import spaces
+
+from scripts.commons.Train_Base import Train_Base
+from scripts.commons.Server import Server as Train_Server
+
+from agent.base_agent import Base_Agent
+from utils.math_ops import MathOps
+
+from scipy.spatial.transform import Rotation as R
+
+'''
+Objective:
+Learn how to run forward using step primitive
+----------
+- class Basic_Run: implements an OpenAI custom gym
+- class Train:  implements algorithms to train a new model or test an existing model
+'''
+
+
+class WalkEnv(gym.Env):
+    def __init__(self, ip, server_p) -> None:
+
+        # Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
+        self.Player = player = Base_Agent(
+            team_name="Gym",
+            number=1,
+            host=ip,
+            port=server_p
+        )
+        self.robot_type = self.Player.robot
+        self.step_counter = 0  # to limit episode size
+        self.force_play_on = True
+
+        self.target_position = np.array([0.0, 0.0])  # target position in the x-y plane
+        self.initial_position = np.array([0.0, 0.0])  # initial position in the x-y plane
+        self.target_direction = 0.0  # target direction in the x-y plane (relative to the robot's orientation)
+        self.isfallen = False
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.debug_every_n_steps = 5
+        self.enable_debug_joint_status = False
+        self.calibrate_nominal_from_neutral = True
+        self.auto_calibrate_train_sim_flip = True
+        self.nominal_calibrated_once = False
+        self.flip_calibrated_once = False
+        self._target_hz = 0.0
+        self._target_dt = 0.0
+        self._last_sync_time = None
+        target_hz_env = 0
+        if target_hz_env:
+            try:
+                self._target_hz = float(target_hz_env)
+            except ValueError:
+                self._target_hz = 0.0
+        if self._target_hz > 0.0:
+            self._target_dt = 1.0 / self._target_hz
+
+        # State space
+        # 原始观测大小: 78
+        obs_size = 78
+        self.obs = np.zeros(obs_size, np.float32)
+        self.observation_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(obs_size,),
+            dtype=np.float32
+        )
+
+        action_dim = len(self.Player.robot.ROBOT_MOTORS)
+        self.no_of_actions = action_dim
+        self.action_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(action_dim,),
+            dtype=np.float32
+        )
+
+        # 中立姿态
+        self.joint_nominal_position = np.array(
+            [
+                0.0,
+                0.0,
+                0.0,
+                1.4,
+                0.0,
+                -0.4,
+                0.0,
+                -1.4,
+                0.0,
+                0.4,
+                0.0,
+                -0.4,
+                0.0,
+                0.0,
+                0.8,
+                -0.4,
+                0.0,
+                0.4,
+                0.0,
+                0.0,
+                -0.8,
+                0.4,
+                0.0,
+            ]
+        )
+        self.joint_nominal_position = np.zeros(self.no_of_actions)
+        self.train_sim_flip = np.array(
+            [
+                1.0,  # 0: Head_yaw (he1)
+                -1.0,  # 1: Head_pitch (he2)
+                1.0,  # 2: Left_Shoulder_Pitch (lae1)
+                -1.0,  # 3: Left_Shoulder_Roll (lae2)
+                -1.0,  # 4: Left_Elbow_Pitch (lae3)
+                1.0,  # 5: Left_Elbow_Yaw (lae4)
+                -1.0,  # 6: Right_Shoulder_Pitch (rae1)
+                -1.0,  # 7: Right_Shoulder_Roll (rae2)
+                1.0,  # 8: Right_Elbow_Pitch (rae3)
+                1.0,  # 9: Right_Elbow_Yaw (rae4)
+                1.0,  # 10: Waist (te1)
+                1.0,  # 11: Left_Hip_Pitch (lle1)
+                -1.0,  # 12: Left_Hip_Roll (lle2)
+                -1.0,  # 13: Left_Hip_Yaw (lle3)
+                1.0,  # 14: Left_Knee_Pitch (lle4)
+                1.0,  # 15: Left_Ankle_Pitch (lle5)
+                -1.0,  # 16: Left_Ankle_Roll (lle6)
+                -1.0,  # 17: Right_Hip_Pitch (rle1)
+                -1.0,  # 18: Right_Hip_Roll (rle2)
+                -1.0,  # 19: Right_Hip_Yaw (rle3)
+                -1.0,  # 20: Right_Knee_Pitch (rle4)
+                -1.0,  # 21: Right_Ankle_Pitch (rle5)
+                -1.0,  # 22: Right_Ankle_Roll (rle6)
+            ]
+        )
+
+        self.scaling_factor = 0.3
+        # self.scaling_factor = 1
+
+        # Encourage a minimum lateral stance so the policy avoids feet overlap.
+        self.min_stance_rad = 0.10
+
+        # Small reset perturbations for robustness training.
+        self.enable_reset_perturb = False
+        self.reset_beam_yaw_range_deg = 180  # randomize target direction fully to encourage learning a real walk instead of a fixed gait
+        self.reset_joint_noise_rad = 0.015
+        self.reset_perturb_steps = 3
+        self.reset_recover_steps = 8
+
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Track previous position
+        self.Player.server.connect()
+        # sleep(2.0)  # Longer wait for connection to establish completely
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+        self.start_time = time.time()
+
+    def _reconnect_server(self):
+        try:
+            self.Player.server.shutdown()
+        except Exception:
+            pass
+
+        self.Player.server.connect()
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+
+    def _safe_receive_world_update(self, retries=1):
+        last_exc = None
+        for attempt in range(retries + 1):
+            try:
+                self.Player.server.receive()
+                self.Player.world.update()
+                return
+            except (ConnectionResetError, OSError) as exc:
+                last_exc = exc
+                if attempt >= retries:
+                    raise
+                self._reconnect_server()
+        if last_exc is not None:
+            raise last_exc
+
+    def debug_log(self, message):
+        print(message)
+        try:
+            log_path = os.path.join(os.path.dirname(os.path.dirname(__file__)), "comm_debug.log")
+            with open(log_path, "a", encoding="utf-8") as f:
+                f.write(message + "\n")
+        except OSError:
+            pass
+
+    def observe(self, init=False):
+
+        """获取当前观测值"""
+        robot = self.Player.robot
+        world = self.Player.world
+
+        # Safety check: ensure data is available
+
+        # 计算目标速度
+        raw_target = self.target_position - world.global_position[:2]
+        velocity = MathOps.rotate_2d_vec(
+            raw_target,
+            -robot.global_orientation_euler[2],
+            is_rad=False
+        )
+
+        # 计算相对方向
+        rel_orientation = MathOps.vector_angle(velocity) * 0.3
+        rel_orientation = np.clip(rel_orientation, -0.25, 0.25)
+
+        velocity = np.concatenate([velocity, np.array([rel_orientation])])
+        velocity[0] = np.clip(velocity[0], -0.5, 0.5)
+        velocity[1] = np.clip(velocity[1], -0.25, 0.25)
+
+        # 关节状态
+        radian_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        radian_joint_speeds = np.deg2rad(
+            [robot.motor_speeds[motor] for motor in robot.ROBOT_MOTORS]
+        )
+
+        qpos_qvel_previous_action = np.concatenate([
+            (radian_joint_positions * self.train_sim_flip - self.joint_nominal_position) / 4.6,
+            radian_joint_speeds / 110.0 * self.train_sim_flip,
+            self.previous_action / 10.0,
+        ])
+
+        # 角速度
+        ang_vel = np.clip(np.deg2rad(robot.gyroscope) / 50.0, -1.0, 1.0)
+
+        # 投影的重力方向
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+
+        # 组合观测
+        observation = np.concatenate([
+            qpos_qvel_previous_action,
+            ang_vel,
+            velocity,
+            projected_gravity,
+        ])
+
+        observation = np.clip(observation, -10.0, 10.0)
+        return observation.astype(np.float32)
+
+    def sync(self):
+        ''' Run a single simulation step '''
+        self._safe_receive_world_update(retries=1)
+        self.Player.robot.commit_motor_targets_pd()
+        self.Player.server.send()
+        if self._target_dt > 0.0:
+            now = time.time()
+            if self._last_sync_time is None:
+                self._last_sync_time = now
+                return
+            elapsed = now - self._last_sync_time
+            remaining = self._target_dt - elapsed
+            if remaining > 0.0:
+                time.sleep(remaining)
+                now = time.time()
+            self._last_sync_time = now
+
+    def debug_joint_status(self):
+        robot = self.Player.robot
+        actual_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        target_joint_positions = getattr(
+            self,
+            'target_joint_positions',
+            np.zeros(len(robot.ROBOT_MOTORS), dtype=np.float32)
+        )
+        joint_error = actual_joint_positions - target_joint_positions
+        leg_slice = slice(11, None)
+
+        self.debug_log(
+            "[WalkDebug] "
+            f"step={self.step_counter} "
+            f"pos={np.round(self.Player.world.global_position, 3).tolist()} "
+            f"target_xy={np.round(self.target_position, 3).tolist()} "
+            f"target_leg={np.round(target_joint_positions[leg_slice], 3).tolist()} "
+            f"actual_leg={np.round(actual_joint_positions[leg_slice], 3).tolist()} "
+            f"err_norm={float(np.linalg.norm(joint_error)):.4f} "
+            f"fallen={self.Player.world.global_position[2] < 0.3}"
+        )
+        print(f"waist target={target_joint_positions[10]:.3f}, actual={actual_joint_positions[10]:.3f}")
+
+    def reset(self, seed=None, options=None):
+        '''
+        Reset and stabilize the robot
+        Note: for some behaviors it would be better to reduce stabilization or add noise
+        '''
+        r = self.Player.robot
+        super().reset(seed=seed)
+        if seed is not None:
+            np.random.seed(seed)
+
+        length1 = 2  # randomize target distance
+        length2 = np.random.uniform(0.6, 1)  # randomize target distance
+        length3 = np.random.uniform(0.6, 1)  # randomize target distance
+        angle2 = np.random.uniform(-30, 30)  # randomize initial orientation
+        angle3 = np.random.uniform(-30, 30)  # randomize target direction
+
+        self.step_counter = 0
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Initialize for first step
+        self.walk_cycle_step = 0
+
+        # 随机 beam 目标位置和朝向，增加训练多样性
+        beam_x = (random() - 0.5) * 10
+        beam_y = (random() - 0.5) * 10
+        beam_yaw = uniform(-self.reset_beam_yaw_range_deg, self.reset_beam_yaw_range_deg)
+
+        for _ in range(5):
+            self._safe_receive_world_update(retries=2)
+            self.Player.robot.commit_motor_targets_pd()
+            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=beam_yaw)
+            self.Player.server.send()
+
+        # 执行 Neutral 技能直到完成，给机器人足够时间在 beam 位置稳定站立
+        finished_count = 0
+        for _ in range(50):
+            finished = self.Player.skills_manager.execute("Neutral")
+            self.sync()
+            if finished:
+                finished_count += 1
+                if finished_count >= 20:  # 假设需要连续20次完成才算成功
+                    break
+
+        if self.enable_reset_perturb and self.reset_joint_noise_rad > 0.0:
+            perturb_action = np.zeros(self.no_of_actions, dtype=np.float32)
+            # Perturb waist + lower body only (10:), keep head/arms stable.
+            perturb_action[10:] = np.random.uniform(
+                -self.reset_joint_noise_rad,
+                self.reset_joint_noise_rad,
+                size=(self.no_of_actions - 10,)
+            )
+
+            for _ in range(self.reset_perturb_steps):
+                target_joint_positions = (self.joint_nominal_position + perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+            for i in range(self.reset_recover_steps):
+                # Linearly fade perturbation to help policy start from near-neutral.
+                alpha = 1.0 - float(i + 1) / float(self.reset_recover_steps)
+                target_joint_positions = (self.joint_nominal_position + alpha * perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+        # memory variables
+        self.sync()  
+        self.initial_position = np.array(self.Player.world.global_position[:2])
+        self.previous_pos = self.initial_position.copy()  # Critical: set to actual position
+        self.act = np.zeros(self.no_of_actions, np.float32)
+        # Build target in the robot's current forward direction instead of fixed global +x.
+        heading_deg = float(r.global_orientation_euler[2])
+        forward_offset = MathOps.rotate_2d_vec(np.array([length1, 0.0]), heading_deg, is_rad=False)
+        point1 = self.initial_position + forward_offset
+        point2 = point1 + MathOps.rotate_2d_vec(np.array([length2, 0]), angle2, is_rad=False)
+        point3 = point2 + MathOps.rotate_2d_vec(np.array([length3, 0]), angle3, is_rad=False)
+        self.point_list = [point1]
+        self.target_position = self.point_list[self.waypoint_index]
+        self.initial_height = self.Player.world.global_position[2]
+
+        return self.observe(True), {}
+
+    def render(self, mode='human', close=False):
+        return
+
+    def compute_reward(self, previous_pos, current_pos, action):
+        height = float(self.Player.world.global_position[2])
+        robot = self.Player.robot
+
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+        tilt_mag = float(np.linalg.norm(projected_gravity[:2]))
+        ang_vel = np.deg2rad(robot.gyroscope)
+        ang_vel_mag = float(np.linalg.norm(ang_vel))
+
+        is_fallen = height < 0.55
+        if is_fallen:
+            # remain = max(0, 800 - self.step_counter)
+            # return -8.0 - 0.01 * remain
+            return -1.0
+        
+
+
+        # # 目标方向
+        # to_target = self.target_position - current_pos
+        # dist_to_target = float(np.linalg.norm(to_target))
+        # if dist_to_target < 0.5:
+        #     return 15.0
+
+        # forward_dir = to_target / dist_to_target if dist_to_target > 0.1 else np.array([1.0, 0.0])
+        # delta_pos = current_pos - previous_pos
+        # forward_step = float(np.dot(delta_pos, forward_dir))
+        # lateral_step = float(np.linalg.norm(delta_pos - forward_dir * forward_step))
+
+        # 奖励项
+        # progress_reward = 2 * forward_step
+        # lateral_penalty = -0.1 * lateral_step
+        alive_bonus = 2.0
+
+        # action_penalty = -0.01 * float(np.linalg.norm(action))
+        smoothness_penalty = -0.01 * float(np.linalg.norm(action - self.last_action_for_reward))
+
+        posture_penalty = -0.3 * (tilt_mag)
+        ang_vel_penalty = -0.02 * ang_vel_mag
+
+        # Use simulator joint readings in training frame to shape lateral stance.
+        joint_pos = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        ) * self.train_sim_flip
+        left_hip_roll = float(joint_pos[12])
+        right_hip_roll = float(joint_pos[18])
+        left_ankle_roll = float(joint_pos[16])
+        right_ankle_roll = float(joint_pos[22])
+
+        hip_spread = left_hip_roll - right_hip_roll
+        ankle_spread = left_ankle_roll - right_ankle_roll
+        stance_metric = 0.6 * abs(hip_spread) + 0.4 * abs(ankle_spread)
+
+        # Penalize narrow stance (feet too close) and scissoring (cross-leg pattern).
+        stance_collapse_penalty = -4.0 * max(0.0, self.min_stance_rad - stance_metric)
+        cross_leg_penalty = -1.2 * max(0.0, -(hip_spread * ankle_spread))
+
+        target_height = self.initial_height
+        height_error =  height - target_height
+        height_penalty = -0.5 * abs(height_error)  # 惩罚高度偏离，系数可调
+
+        # # 在 compute_reward 开头附近，添加高度变化率计算
+        # if not hasattr(self, 'last_height'):
+        #     self.last_height = height
+        #     self.last_height_time = self.step_counter  # 可选，用于时间间隔
+        # height_rate = height - self.last_height  # 正为上升，负为下降
+        # self.last_height = height
+
+        # 惩罚高度下降（负变化率）
+        # height_down_penalty = -5.0 * max(0, -height_rate)  # 系数可调，-height_rate 为正表示下降幅度
+
+        # # 在 compute_reward 中
+        # if self.step_counter > 50:
+        #     avg_prev_action = np.mean(self.prev_action_history, axis=0)
+        #     novelty = float(np.linalg.norm(action - avg_prev_action))
+        #     exploration_bonus = 0.05 * novelty
+        # else:
+        #     exploration_bonus = 0
+
+        # self.prev_action_history[self.history_idx] = action
+        # self.history_idx = (self.history_idx + 1) % 50
+
+
+        total = (
+            # progress_reward  + 
+                 alive_bonus + 
+                 # lateral_penalty +
+                 # action_penalty +
+                smoothness_penalty +
+                 posture_penalty
+                 + ang_vel_penalty
+                 + height_penalty
+                 + stance_collapse_penalty
+                 + cross_leg_penalty
+                #  + exploration_bonus
+                # + height_down_penalty
+                 )
+        if time.time() - self.start_time >= 600:
+            self.start_time = time.time()
+            print(
+                #   f"progress_reward:{progress_reward:.4f}",
+                #   f"lateral_penalty:{lateral_penalty:.4f}",
+                #   f"action_penalty:{action_penalty:.4f}"s, 
+                  f"height_penalty:{height_penalty:.4f}",
+                  f"smoothness_penalty:{smoothness_penalty:.4f},",
+                  f"posture_penalty:{posture_penalty:.4f}",
+                  f"stance_collapse_penalty:{stance_collapse_penalty:.4f}",
+                  f"cross_leg_penalty:{cross_leg_penalty:.4f}",
+                #   f"ang_vel_penalty:{ang_vel_penalty:.4f}",
+                #   f"height_down_penalty:{height_down_penalty:.4f}",
+                #   f"exploration_bonus:{exploration_bonus:.4f}"
+            )
+
+        return total
+
+
+
+    def step(self, action):
+
+        r = self.Player.robot
+        self.previous_action = action
+
+        self.target_joint_positions = (
+                # self.joint_nominal_position +
+                 self.scaling_factor * action
+        )
+        self.target_joint_positions *= self.train_sim_flip
+
+        for idx, target in enumerate(self.target_joint_positions):
+            r.set_motor_target_position(
+                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=40, kd=1.0
+            )
+
+        self.previous_action = action
+
+        self.sync()  # run simulation step
+        self.step_counter += 1
+
+        if self.enable_debug_joint_status and self.step_counter % self.debug_every_n_steps == 0:
+            self.debug_joint_status()
+
+        current_pos = np.array(self.Player.world.global_position[:2], dtype=np.float32)
+
+        # Compute reward based on movement from previous step
+        reward = self.compute_reward(self.previous_pos, current_pos, action)
+
+        # Update previous position
+        self.previous_pos = current_pos.copy()
+        self.last_action_for_reward = action.copy()
+
+        # Fall detection and penalty
+        is_fallen = self.Player.world.global_position[2] < 0.55
+
+        # terminal state: the robot is falling or timeout
+        terminated = is_fallen or self.step_counter > 800 or self.route_completed
+        truncated = False
+
+        return self.observe(), reward, terminated, truncated, {}
+
+
+class Train(Train_Base):
+    def __init__(self, script) -> None:
+        super().__init__(script)
+
+    def train(self, args):
+
+        # --------------------------------------- Learning parameters
+        n_envs = int(os.environ.get("GYM_CPU_N_ENVS", "20"))
+        if n_envs < 1:
+            raise ValueError("GYM_CPU_N_ENVS must be >= 1")
+        server_warmup_sec = float(os.environ.get("GYM_CPU_SERVER_WARMUP_SEC", "3.0"))
+        n_steps_per_env = int(os.environ.get("GYM_CPU_TRAIN_STEPS_PER_ENV", "256"))  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        minibatch_size = int(os.environ.get("GYM_CPU_TRAIN_BATCH_SIZE", "512"))  # should be a factor of (n_steps_per_env * n_envs)
+        total_steps = 30000000
+        learning_rate = float(os.environ.get("GYM_CPU_TRAIN_LR", "3e-4"))
+        folder_name = f'Walk_R{self.robot_type}'
+        model_path = f'./scripts/gyms/logs/{folder_name}/'
+
+        print(f"Model path: {model_path}")
+        print(f"Using {n_envs} parallel environments")
+
+        # --------------------------------------- Run algorithm
+        def init_env(i_env, monitor=False):
+            def thunk():
+                env = WalkEnv(self.ip, self.server_p + i_env)
+                if monitor:
+                    env = Monitor(env)
+                return env
+
+            return thunk
+
+        server_log_dir = os.path.join(model_path, "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1, no_render=True, no_realtime=True)  # include 1 extra server for testing
+
+        # Wait for servers to start
+        print(f"Starting {n_envs + 1} rcssservermj servers...")
+        if server_warmup_sec > 0:
+            print(f"Waiting {server_warmup_sec:.1f}s for server warmup...")
+            sleep(server_warmup_sec)
+        print("Servers started, creating environments...")
+
+        env = SubprocVecEnv([init_env(i, monitor=True) for i in range(n_envs)])
+        # Use single-process eval env to avoid extra subprocess fragility during callback evaluation.
+        eval_env = DummyVecEnv([init_env(n_envs, monitor=True)])
+
+        try:
+            # Custom policy network architecture
+            policy_kwargs = dict(
+                net_arch=dict(
+                    pi=[512, 256, 128],  # Policy network: 3 layers
+                    vf=[512, 256, 128]  # Value network: 3 layers
+                ),
+                activation_fn=__import__('torch.nn', fromlist=['ELU']).ELU,
+            )
+
+            if "model_file" in args:  # retrain
+                model = PPO.load(args["model_file"], env=env, device="cpu", n_envs=n_envs, n_steps=n_steps_per_env,
+                                 batch_size=minibatch_size, learning_rate=learning_rate)
+            else:  # train new model
+                model = PPO(
+                    "MlpPolicy",
+                    env=env,
+                    verbose=1,
+                    n_steps=n_steps_per_env,
+                    batch_size=minibatch_size,
+                    learning_rate=learning_rate,
+                    device="cpu",
+                    policy_kwargs=policy_kwargs,
+                    ent_coef=float(os.environ.get("GYM_CPU_TRAIN_ENT_COEF", "0.05")),  # Entropy coefficient for exploration
+                    clip_range=float(os.environ.get("GYM_CPU_TRAIN_CLIP_RANGE", "0.2")),  # PPO clipping parameter
+                    gae_lambda=0.95,  # GAE lambda
+                    gamma=float(os.environ.get("GYM_CPU_TRAIN_GAMMA", "0.95")), # Discount factor
+                    # target_kl=0.03,
+                    n_epochs=int(os.environ.get("GYM_CPU_TRAIN_EPOCHS", "5")),
+                    tensorboard_log=f"./scripts/gyms/logs/{folder_name}/tensorboard/"
+                )
+
+            model_path = self.learn_model(model, total_steps, model_path, eval_env=eval_env,
+                                          eval_freq=n_steps_per_env * 20, save_freq=n_steps_per_env * 20, eval_eps=100,
+                                          backup_env_file=__file__)
+        except KeyboardInterrupt:
+            sleep(1)  # wait for child processes
+            print("\nctrl+c pressed, aborting...\n")
+            servers.kill()
+            return
+
+        env.close()
+        eval_env.close()
+        servers.kill()
+
+    def test(self, args):
+
+        # Uses different server and monitor ports
+        server_log_dir = os.path.join(args["folder_dir"], "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        test_no_render = os.environ.get("GYM_CPU_TEST_NO_RENDER", "0") == "1"
+        test_no_realtime = os.environ.get("GYM_CPU_TEST_NO_REALTIME", "0") == "1"
+
+        server = Train_Server(
+            self.server_p - 1,
+            self.monitor_p,
+            1,
+            no_render=test_no_render,
+            no_realtime=test_no_realtime,
+        )
+        env = WalkEnv(self.ip, self.server_p - 1)
+        model = PPO.load(args["model_file"], env=env)
+
+        try:
+            self.export_model(args["model_file"], args["model_file"] + ".pkl",
+                              False)  # Export to pkl to create custom behavior
+            self.test_model(model, env, log_path=args["folder_dir"], model_path=args["folder_dir"])
+        except KeyboardInterrupt:
+            print()
+
+        env.close()
+        server.kill()
+
+
+if __name__ == "__main__":
+    from types import SimpleNamespace
+
+    # 创建默认参数
+    script_args = SimpleNamespace(
+        args=SimpleNamespace(
+            i='127.0.0.1',  # Server IP
+            p=3100,  # Server port
+            m=3200,  # Monitor port
+            r=0,  # Robot type
+            t='Gym',  # Team name
+            u=1  # Uniform number
+        )
+    )
+
+    trainer = Train(script_args)
+
+    run_mode = os.environ.get("GYM_CPU_MODE", "train").strip().lower()
+
+    if run_mode == "test":
+        test_model_file = os.environ.get("GYM_CPU_TEST_MODEL", "scripts/gyms/logs/Walk_R0_004/best_model.zip")
+        test_folder = os.environ.get("GYM_CPU_TEST_FOLDER", "scripts/gyms/logs/Walk_R0_004/")
+        trainer.test({"model_file": test_model_file, "folder_dir": test_folder})
+    else:
+        retrain_model = os.environ.get("GYM_CPU_TRAIN_MODEL", "").strip()
+        if retrain_model:
+            trainer.train({"model_file": retrain_model})
+        else:
+            trainer.train({})

scripts/gyms/logs/stand_stable_0.1/Walk.py

@@ -0,0 +1,626 @@
+import os
+import numpy as np
+import math
+import time
+from time import sleep
+from random import random
+from random import uniform
+
+from stable_baselines3 import PPO
+from stable_baselines3.common.vec_env import SubprocVecEnv
+
+import gymnasium as gym
+from gymnasium import spaces
+
+from scripts.commons.Train_Base import Train_Base
+from scripts.commons.Server import Server as Train_Server
+
+from agent.base_agent import Base_Agent
+from utils.math_ops import MathOps
+
+from scipy.spatial.transform import Rotation as R
+
+'''
+Objective:
+Learn how to run forward using step primitive
+----------
+- class Basic_Run: implements an OpenAI custom gym
+- class Train:  implements algorithms to train a new model or test an existing model
+'''
+
+
+class WalkEnv(gym.Env):
+    def __init__(self, ip, server_p) -> None:
+
+        # Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
+        self.Player = player = Base_Agent(
+            team_name="Gym",
+            number=1,
+            host=ip,
+            port=server_p
+        )
+        self.robot_type = self.Player.robot
+        self.step_counter = 0  # to limit episode size
+        self.force_play_on = True
+
+        self.target_position = np.array([0.0, 0.0])  # target position in the x-y plane
+        self.initial_position = np.array([0.0, 0.0])  # initial position in the x-y plane
+        self.target_direction = 0.0  # target direction in the x-y plane (relative to the robot's orientation)
+        self.isfallen = False
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.debug_every_n_steps = 5
+        self.enable_debug_joint_status = False
+        self.calibrate_nominal_from_neutral = True
+        self.auto_calibrate_train_sim_flip = True
+        self.nominal_calibrated_once = False
+        self.flip_calibrated_once = False
+        self._target_hz = 0.0
+        self._target_dt = 0.0
+        self._last_sync_time = None
+        target_hz_env = 0
+        if target_hz_env:
+            try:
+                self._target_hz = float(target_hz_env)
+            except ValueError:
+                self._target_hz = 0.0
+        if self._target_hz > 0.0:
+            self._target_dt = 1.0 / self._target_hz
+
+        # State space
+        # 原始观测大小: 78
+        obs_size = 78
+        self.obs = np.zeros(obs_size, np.float32)
+        self.observation_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(obs_size,),
+            dtype=np.float32
+        )
+
+        action_dim = len(self.Player.robot.ROBOT_MOTORS)
+        self.no_of_actions = action_dim
+        self.action_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(action_dim,),
+            dtype=np.float32
+        )
+
+        # 中立姿态
+        self.joint_nominal_position = np.array(
+            [
+                0.0,
+                0.0,
+                0.0,
+                1.4,
+                0.0,
+                -0.4,
+                0.0,
+                -1.4,
+                0.0,
+                0.4,
+                0.0,
+                -0.4,
+                0.0,
+                0.0,
+                0.8,
+                -0.4,
+                0.0,
+                0.4,
+                0.0,
+                0.0,
+                -0.8,
+                0.4,
+                0.0,
+            ]
+        )
+        self.joint_nominal_position = np.zeros(self.no_of_actions)
+        self.train_sim_flip = np.array(
+            [
+                1.0,  # 0: Head_yaw (he1)
+                -1.0,  # 1: Head_pitch (he2)
+                1.0,  # 2: Left_Shoulder_Pitch (lae1)
+                -1.0,  # 3: Left_Shoulder_Roll (lae2)
+                -1.0,  # 4: Left_Elbow_Pitch (lae3)
+                1.0,  # 5: Left_Elbow_Yaw (lae4)
+                -1.0,  # 6: Right_Shoulder_Pitch (rae1)
+                -1.0,  # 7: Right_Shoulder_Roll (rae2)
+                1.0,  # 8: Right_Elbow_Pitch (rae3)
+                1.0,  # 9: Right_Elbow_Yaw (rae4)
+                1.0,  # 10: Waist (te1)
+                1.0,  # 11: Left_Hip_Pitch (lle1)
+                -1.0,  # 12: Left_Hip_Roll (lle2)
+                -1.0,  # 13: Left_Hip_Yaw (lle3)
+                1.0,  # 14: Left_Knee_Pitch (lle4)
+                1.0,  # 15: Left_Ankle_Pitch (lle5)
+                -1.0,  # 16: Left_Ankle_Roll (lle6)
+                -1.0,  # 17: Right_Hip_Pitch (rle1)
+                -1.0,  # 18: Right_Hip_Roll (rle2)
+                -1.0,  # 19: Right_Hip_Yaw (rle3)
+                -1.0,  # 20: Right_Knee_Pitch (rle4)
+                -1.0,  # 21: Right_Ankle_Pitch (rle5)
+                -1.0,  # 22: Right_Ankle_Roll (rle6)
+            ]
+        )
+
+        self.scaling_factor = 0.3
+        # self.scaling_factor = 1
+
+        # Small reset perturbations for robustness training.
+        self.enable_reset_perturb = True
+        self.reset_beam_yaw_range_deg = 180  # randomize target direction fully to encourage learning a real walk instead of a fixed gait
+        self.reset_joint_noise_rad = 0.015
+        self.reset_perturb_steps = 3
+        self.reset_recover_steps = 8
+
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Track previous position
+        self.Player.server.connect()
+        # sleep(2.0)  # Longer wait for connection to establish completely
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+        self.start_time = time.time()
+
+    def debug_log(self, message):
+        print(message)
+        try:
+            log_path = os.path.join(os.path.dirname(os.path.dirname(__file__)), "comm_debug.log")
+            with open(log_path, "a", encoding="utf-8") as f:
+                f.write(message + "\n")
+        except OSError:
+            pass
+
+    def observe(self, init=False):
+
+        """获取当前观测值"""
+        robot = self.Player.robot
+        world = self.Player.world
+
+        # Safety check: ensure data is available
+
+        # 计算目标速度
+        raw_target = self.target_position - world.global_position[:2]
+        velocity = MathOps.rotate_2d_vec(
+            raw_target,
+            -robot.global_orientation_euler[2],
+            is_rad=False
+        )
+
+        # 计算相对方向
+        rel_orientation = MathOps.vector_angle(velocity) * 0.3
+        rel_orientation = np.clip(rel_orientation, -0.25, 0.25)
+
+        velocity = np.concatenate([velocity, np.array([rel_orientation])])
+        velocity[0] = np.clip(velocity[0], -0.5, 0.5)
+        velocity[1] = np.clip(velocity[1], -0.25, 0.25)
+
+        # 关节状态
+        radian_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        radian_joint_speeds = np.deg2rad(
+            [robot.motor_speeds[motor] for motor in robot.ROBOT_MOTORS]
+        )
+
+        qpos_qvel_previous_action = np.concatenate([
+            (radian_joint_positions * self.train_sim_flip - self.joint_nominal_position) / 4.6,
+            radian_joint_speeds / 110.0 * self.train_sim_flip,
+            self.previous_action / 10.0,
+        ])
+
+        # 角速度
+        ang_vel = np.clip(np.deg2rad(robot.gyroscope) / 50.0, -1.0, 1.0)
+
+        # 投影的重力方向
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+
+        # 组合观测
+        observation = np.concatenate([
+            qpos_qvel_previous_action,
+            ang_vel,
+            velocity,
+            projected_gravity,
+        ])
+
+        observation = np.clip(observation, -10.0, 10.0)
+        return observation.astype(np.float32)
+
+    def sync(self):
+        ''' Run a single simulation step '''
+        self.Player.server.receive()
+        self.Player.world.update()
+        self.Player.robot.commit_motor_targets_pd()
+        self.Player.server.send()
+        if self._target_dt > 0.0:
+            now = time.time()
+            if self._last_sync_time is None:
+                self._last_sync_time = now
+                return
+            elapsed = now - self._last_sync_time
+            remaining = self._target_dt - elapsed
+            if remaining > 0.0:
+                time.sleep(remaining)
+                now = time.time()
+            self._last_sync_time = now
+
+    def debug_joint_status(self):
+        robot = self.Player.robot
+        actual_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        target_joint_positions = getattr(
+            self,
+            'target_joint_positions',
+            np.zeros(len(robot.ROBOT_MOTORS), dtype=np.float32)
+        )
+        joint_error = actual_joint_positions - target_joint_positions
+        leg_slice = slice(11, None)
+
+        self.debug_log(
+            "[WalkDebug] "
+            f"step={self.step_counter} "
+            f"pos={np.round(self.Player.world.global_position, 3).tolist()} "
+            f"target_xy={np.round(self.target_position, 3).tolist()} "
+            f"target_leg={np.round(target_joint_positions[leg_slice], 3).tolist()} "
+            f"actual_leg={np.round(actual_joint_positions[leg_slice], 3).tolist()} "
+            f"err_norm={float(np.linalg.norm(joint_error)):.4f} "
+            f"fallen={self.Player.world.global_position[2] < 0.3}"
+        )
+        print(f"waist target={target_joint_positions[10]:.3f}, actual={actual_joint_positions[10]:.3f}")
+
+    def reset(self, seed=None, options=None):
+        '''
+        Reset and stabilize the robot
+        Note: for some behaviors it would be better to reduce stabilization or add noise
+        '''
+        r = self.Player.robot
+        super().reset(seed=seed)
+        if seed is not None:
+            np.random.seed(seed)
+
+        length1 = 2  # randomize target distance
+        length2 = np.random.uniform(0.6, 1)  # randomize target distance
+        length3 = np.random.uniform(0.6, 1)  # randomize target distance
+        angle2 = np.random.uniform(-30, 30)  # randomize initial orientation
+        angle3 = np.random.uniform(-30, 30)  # randomize target direction
+
+        self.step_counter = 0
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Initialize for first step
+        self.walk_cycle_step = 0
+
+        # 随机 beam 目标位置和朝向，增加训练多样性
+        beam_x = (random() - 0.5) * 10
+        beam_y = (random() - 0.5) * 10
+        beam_yaw = uniform(-self.reset_beam_yaw_range_deg, self.reset_beam_yaw_range_deg)
+
+        for _ in range(5):
+            self.Player.server.receive()
+            self.Player.world.update()
+            self.Player.robot.commit_motor_targets_pd()
+            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=beam_yaw)
+            self.Player.server.send()
+
+        # 执行 Neutral 技能直到完成，给机器人足够时间在 beam 位置稳定站立
+        finished_count = 0
+        for _ in range(50):
+            finished = self.Player.skills_manager.execute("Neutral")
+            self.sync()
+            if finished:
+                finished_count += 1
+                if finished_count >= 20:  # 假设需要连续20次完成才算成功
+                    break
+
+        if self.enable_reset_perturb and self.reset_joint_noise_rad > 0.0:
+            perturb_action = np.zeros(self.no_of_actions, dtype=np.float32)
+            # Perturb waist + lower body only (10:), keep head/arms stable.
+            perturb_action[10:] = np.random.uniform(
+                -self.reset_joint_noise_rad,
+                self.reset_joint_noise_rad,
+                size=(self.no_of_actions - 10,)
+            )
+
+            for _ in range(self.reset_perturb_steps):
+                target_joint_positions = (self.joint_nominal_position + perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+            for i in range(self.reset_recover_steps):
+                # Linearly fade perturbation to help policy start from near-neutral.
+                alpha = 1.0 - float(i + 1) / float(self.reset_recover_steps)
+                target_joint_positions = (self.joint_nominal_position + alpha * perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+        # memory variables
+        self.sync()  
+        self.initial_position = np.array(self.Player.world.global_position[:2])
+        self.previous_pos = self.initial_position.copy()  # Critical: set to actual position
+        self.act = np.zeros(self.no_of_actions, np.float32)
+        # Build target in the robot's current forward direction instead of fixed global +x.
+        heading_deg = float(r.global_orientation_euler[2])
+        forward_offset = MathOps.rotate_2d_vec(np.array([length1, 0.0]), heading_deg, is_rad=False)
+        point1 = self.initial_position + forward_offset
+        point2 = point1 + MathOps.rotate_2d_vec(np.array([length2, 0]), angle2, is_rad=False)
+        point3 = point2 + MathOps.rotate_2d_vec(np.array([length3, 0]), angle3, is_rad=False)
+        self.point_list = [point1]
+        self.target_position = self.point_list[self.waypoint_index]
+        self.initial_height = self.Player.world.global_position[2]
+
+        return self.observe(True), {}
+
+    def render(self, mode='human', close=False):
+        return
+
+    def compute_reward(self, previous_pos, current_pos, action):
+        height = float(self.Player.world.global_position[2])
+
+        orientation_quat_inv = R.from_quat(self.Player.robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+        tilt_mag = float(np.linalg.norm(projected_gravity[:2]))
+        ang_vel = np.deg2rad(self.Player.robot.gyroscope)
+        ang_vel_mag = float(np.linalg.norm(ang_vel))
+
+        is_fallen = height < 0.3
+        if is_fallen:
+            # remain = max(0, 800 - self.step_counter)
+            # return -8.0 - 0.01 * remain
+            return -1.0
+        
+
+
+        # # 目标方向
+        # to_target = self.target_position - current_pos
+        # dist_to_target = float(np.linalg.norm(to_target))
+        # if dist_to_target < 0.5:
+        #     return 15.0
+
+        # forward_dir = to_target / dist_to_target if dist_to_target > 0.1 else np.array([1.0, 0.0])
+        # delta_pos = current_pos - previous_pos
+        # forward_step = float(np.dot(delta_pos, forward_dir))
+        # lateral_step = float(np.linalg.norm(delta_pos - forward_dir * forward_step))
+
+        # 奖励项
+        # progress_reward = 2 * forward_step
+        # lateral_penalty = -0.1 * lateral_step
+        alive_bonus = 2.0
+
+        # action_penalty = -0.01 * float(np.linalg.norm(action))
+        smoothness_penalty = -0.01 * float(np.linalg.norm(action - self.last_action_for_reward))
+
+        posture_penalty = -0.3 * (tilt_mag)
+        ang_vel_penalty = -0.02 * ang_vel_mag
+
+        target_height = self.initial_height
+        height_error =  height - target_height
+        height_penalty = -0.5 * abs(height_error)  # 惩罚高度偏离，系数可调
+
+        # # 在 compute_reward 开头附近，添加高度变化率计算
+        # if not hasattr(self, 'last_height'):
+        #     self.last_height = height
+        #     self.last_height_time = self.step_counter  # 可选，用于时间间隔
+        # height_rate = height - self.last_height  # 正为上升，负为下降
+        # self.last_height = height
+
+        # 惩罚高度下降（负变化率）
+        # height_down_penalty = -5.0 * max(0, -height_rate)  # 系数可调，-height_rate 为正表示下降幅度
+
+        # # 在 compute_reward 中
+        # if self.step_counter > 50:
+        #     avg_prev_action = np.mean(self.prev_action_history, axis=0)
+        #     novelty = float(np.linalg.norm(action - avg_prev_action))
+        #     exploration_bonus = 0.05 * novelty
+        # else:
+        #     exploration_bonus = 0
+
+        # self.prev_action_history[self.history_idx] = action
+        # self.history_idx = (self.history_idx + 1) % 50
+
+
+        total = (
+            # progress_reward  + 
+                 alive_bonus + 
+                 # lateral_penalty +
+                 # action_penalty +
+                smoothness_penalty +
+                 posture_penalty
+                 + ang_vel_penalty
+                 + height_penalty
+                #  + exploration_bonus
+                # + height_down_penalty
+                 )
+        if time.time() - self.start_time >= 1200:
+            self.start_time = time.time()
+            print(
+                #   f"progress_reward:{progress_reward:.4f}",
+                #   f"lateral_penalty:{lateral_penalty:.4f}",
+                #   f"action_penalty:{action_penalty:.4f}"s, 
+                  f"height_penalty:{height_penalty:.4f}",
+                  f"smoothness_penalty:{smoothness_penalty:.4f},",
+                  f"posture_penalty:{posture_penalty:.4f}",
+                #   f"ang_vel_penalty:{ang_vel_penalty:.4f}",
+                #   f"height_down_penalty:{height_down_penalty:.4f}",
+                #   f"exploration_bonus:{exploration_bonus:.4f}"
+            )
+
+        return total
+
+
+
+    def step(self, action):
+
+        r = self.Player.robot
+        self.previous_action = action
+
+        self.target_joint_positions = (
+                # self.joint_nominal_position +
+                 self.scaling_factor * action
+        )
+        self.target_joint_positions *= self.train_sim_flip
+
+        for idx, target in enumerate(self.target_joint_positions):
+            r.set_motor_target_position(
+                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=40, kd=1.0
+            )
+
+        self.previous_action = action
+
+        self.sync()  # run simulation step
+        self.step_counter += 1
+
+        if self.enable_debug_joint_status and self.step_counter % self.debug_every_n_steps == 0:
+            self.debug_joint_status()
+
+        current_pos = np.array(self.Player.world.global_position[:2], dtype=np.float32)
+
+        # Compute reward based on movement from previous step
+        reward = self.compute_reward(self.previous_pos, current_pos, action)
+
+        # Update previous position
+        self.previous_pos = current_pos.copy()
+        self.last_action_for_reward = action.copy()
+
+        # Fall detection and penalty
+        is_fallen = self.Player.world.global_position[2] < 0.3
+
+        # terminal state: the robot is falling or timeout
+        terminated = is_fallen or self.step_counter > 800 or self.route_completed
+        truncated = False
+
+        return self.observe(), reward, terminated, truncated, {}
+
+
+class Train(Train_Base):
+    def __init__(self, script) -> None:
+        super().__init__(script)
+
+    def train(self, args):
+
+        # --------------------------------------- Learning parameters
+        n_envs = 20  # Reduced from 8 to decrease CPU/network pressure during init
+        if n_envs < 1:
+            raise ValueError("GYM_CPU_N_ENVS must be >= 1")
+        n_steps_per_env = 256  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        minibatch_size = 512  # should be a factor of (n_steps_per_env * n_envs)
+        total_steps = 30000000
+        learning_rate = 3e-4
+        folder_name = f'Walk_R{self.robot_type}'
+        model_path = f'./scripts/gyms/logs/{folder_name}/'
+
+        print(f"Model path: {model_path}")
+        print(f"Using {n_envs} parallel environments")
+
+        # --------------------------------------- Run algorithm
+        def init_env(i_env):
+            def thunk():
+                return WalkEnv(self.ip, self.server_p + i_env)
+
+            return thunk
+
+        server_log_dir = os.path.join(model_path, "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1)  # include 1 extra server for testing
+
+        # Wait for servers to start
+        print(f"Starting {n_envs + 1} rcssservermj servers...")
+        print("Servers started, creating environments...")
+
+        env = SubprocVecEnv([init_env(i) for i in range(n_envs)])
+        eval_env = SubprocVecEnv([init_env(n_envs)])
+
+        try:
+            # Custom policy network architecture
+            policy_kwargs = dict(
+                net_arch=dict(
+                    pi=[512, 256, 128],  # Policy network: 3 layers
+                    vf=[512, 256, 128]  # Value network: 3 layers
+                ),
+                activation_fn=__import__('torch.nn', fromlist=['ELU']).ELU,
+            )
+
+            if "model_file" in args:  # retrain
+                model = PPO.load(args["model_file"], env=env, device="cpu", n_envs=n_envs, n_steps=n_steps_per_env,
+                                 batch_size=minibatch_size, learning_rate=learning_rate)
+            else:  # train new model
+                model = PPO(
+                    "MlpPolicy",
+                    env=env,
+                    verbose=1,
+                    n_steps=n_steps_per_env,
+                    batch_size=minibatch_size,
+                    learning_rate=learning_rate,
+                    device="cpu",
+                    policy_kwargs=policy_kwargs,
+                    ent_coef=0.05,  # Entropy coefficient for exploration
+                    # clip_range=0.13,  # PPO clipping parameter
+                    gae_lambda=0.95,  # GAE lambda
+                    gamma=0.95 , # Discount factor
+                    target_kl=0.03,
+                    n_epochs=5
+                )
+
+            model_path = self.learn_model(model, total_steps, model_path, eval_env=eval_env,
+                                          eval_freq=n_steps_per_env * 10, save_freq=n_steps_per_env * 10,
+                                          backup_env_file=__file__)
+        except KeyboardInterrupt:
+            sleep(1)  # wait for child processes
+            print("\nctrl+c pressed, aborting...\n")
+            servers.kill()
+            return
+
+        env.close()
+        eval_env.close()
+        servers.kill()
+
+    def test(self, args):
+
+        # Uses different server and monitor ports
+        server_log_dir = os.path.join(args["folder_dir"], "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        server = Train_Server(self.server_p - 1, self.monitor_p, 1)
+        env = WalkEnv(self.ip, self.server_p - 1)
+        model = PPO.load(args["model_file"], env=env)
+
+        try:
+            self.export_model(args["model_file"], args["model_file"] + ".pkl",
+                              False)  # Export to pkl to create custom behavior
+            self.test_model(model, env, log_path=args["folder_dir"], model_path=args["folder_dir"])
+        except KeyboardInterrupt:
+            print()
+
+        env.close()
+        server.kill()
+
+
+if __name__ == "__main__":
+    from types import SimpleNamespace
+
+    # 创建默认参数
+    script_args = SimpleNamespace(
+        args=SimpleNamespace(
+            i='127.0.0.1',  # Server IP
+            p=3100,  # Server port
+            m=3200,  # Monitor port
+            r=0,  # Robot type
+            t='Gym',  # Team name
+            u=1  # Uniform number
+        )
+    )
+
+    trainer = Train(script_args)
+    trainer.train({})
+    # trainer.test({"model_file": "scripts/gyms/logs/Walk_R0_000/best_model.zip",
+                #   "folder_dir": "scripts/gyms/logs/Walk_R0_000/",})

scripts/gyms/logs/stand_stable_final/Walk.py

@@ -0,0 +1,705 @@
+import os
+import numpy as np
+import math
+import time
+from time import sleep
+from random import random
+from random import uniform
+from itertools import count
+
+from stable_baselines3 import PPO
+from stable_baselines3.common.monitor import Monitor
+from stable_baselines3.common.vec_env import SubprocVecEnv, DummyVecEnv
+
+import gymnasium as gym
+from gymnasium import spaces
+
+from scripts.commons.Train_Base import Train_Base
+from scripts.commons.Server import Server as Train_Server
+
+from agent.base_agent import Base_Agent
+from utils.math_ops import MathOps
+
+from scipy.spatial.transform import Rotation as R
+
+'''
+Objective:
+Learn how to run forward using step primitive
+----------
+- class Basic_Run: implements an OpenAI custom gym
+- class Train:  implements algorithms to train a new model or test an existing model
+'''
+
+
+class WalkEnv(gym.Env):
+    def __init__(self, ip, server_p) -> None:
+
+        # Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
+        self.Player = player = Base_Agent(
+            team_name="Gym",
+            number=1,
+            host=ip,
+            port=server_p
+        )
+        self.robot_type = self.Player.robot
+        self.step_counter = 0  # to limit episode size
+        self.force_play_on = True
+
+        self.target_position = np.array([0.0, 0.0])  # target position in the x-y plane
+        self.initial_position = np.array([0.0, 0.0])  # initial position in the x-y plane
+        self.target_direction = 0.0  # target direction in the x-y plane (relative to the robot's orientation)
+        self.isfallen = False
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.debug_every_n_steps = 5
+        self.enable_debug_joint_status = False
+        self.calibrate_nominal_from_neutral = True
+        self.auto_calibrate_train_sim_flip = True
+        self.nominal_calibrated_once = False
+        self.flip_calibrated_once = False
+        self._target_hz = 0.0
+        self._target_dt = 0.0
+        self._last_sync_time = None
+        target_hz_env = 0
+        if target_hz_env:
+            try:
+                self._target_hz = float(target_hz_env)
+            except ValueError:
+                self._target_hz = 0.0
+        if self._target_hz > 0.0:
+            self._target_dt = 1.0 / self._target_hz
+
+        # State space
+        # 原始观测大小: 78
+        obs_size = 78
+        self.obs = np.zeros(obs_size, np.float32)
+        self.observation_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(obs_size,),
+            dtype=np.float32
+        )
+
+        action_dim = len(self.Player.robot.ROBOT_MOTORS)
+        self.no_of_actions = action_dim
+        self.action_space = spaces.Box(
+            low=-10.0,
+            high=10.0,
+            shape=(action_dim,),
+            dtype=np.float32
+        )
+
+        # 中立姿态
+        self.joint_nominal_position = np.array(
+            [
+                0.0,
+                0.0,
+                0.0,
+                1.4,
+                0.0,
+                -0.4,
+                0.0,
+                -1.4,
+                0.0,
+                0.4,
+                0.0,
+                -0.4,
+                0.0,
+                0.0,
+                0.8,
+                -0.4,
+                0.0,
+                0.4,
+                0.0,
+                0.0,
+                -0.8,
+                0.4,
+                0.0,
+            ]
+        )
+        self.joint_nominal_position = np.zeros(self.no_of_actions)
+        self.train_sim_flip = np.array(
+            [
+                1.0,  # 0: Head_yaw (he1)
+                -1.0,  # 1: Head_pitch (he2)
+                1.0,  # 2: Left_Shoulder_Pitch (lae1)
+                -1.0,  # 3: Left_Shoulder_Roll (lae2)
+                -1.0,  # 4: Left_Elbow_Pitch (lae3)
+                1.0,  # 5: Left_Elbow_Yaw (lae4)
+                -1.0,  # 6: Right_Shoulder_Pitch (rae1)
+                -1.0,  # 7: Right_Shoulder_Roll (rae2)
+                1.0,  # 8: Right_Elbow_Pitch (rae3)
+                1.0,  # 9: Right_Elbow_Yaw (rae4)
+                1.0,  # 10: Waist (te1)
+                1.0,  # 11: Left_Hip_Pitch (lle1)
+                -1.0,  # 12: Left_Hip_Roll (lle2)
+                -1.0,  # 13: Left_Hip_Yaw (lle3)
+                1.0,  # 14: Left_Knee_Pitch (lle4)
+                1.0,  # 15: Left_Ankle_Pitch (lle5)
+                -1.0,  # 16: Left_Ankle_Roll (lle6)
+                -1.0,  # 17: Right_Hip_Pitch (rle1)
+                -1.0,  # 18: Right_Hip_Roll (rle2)
+                -1.0,  # 19: Right_Hip_Yaw (rle3)
+                -1.0,  # 20: Right_Knee_Pitch (rle4)
+                -1.0,  # 21: Right_Ankle_Pitch (rle5)
+                -1.0,  # 22: Right_Ankle_Roll (rle6)
+            ]
+        )
+
+        self.scaling_factor = 0.3
+        # self.scaling_factor = 1
+
+        # Encourage a minimum lateral stance so the policy avoids feet overlap.
+        self.min_stance_rad = 0.10
+
+        # Small reset perturbations for robustness training.
+        self.enable_reset_perturb = False
+        self.reset_beam_yaw_range_deg = 180  # randomize target direction fully to encourage learning a real walk instead of a fixed gait
+        self.reset_joint_noise_rad = 0.015
+        self.reset_perturb_steps = 3
+        self.reset_recover_steps = 8
+
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Track previous position
+        self.Player.server.connect()
+        # sleep(2.0)  # Longer wait for connection to establish completely
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+        self.start_time = time.time()
+
+    def _reconnect_server(self):
+        try:
+            self.Player.server.shutdown()
+        except Exception:
+            pass
+
+        self.Player.server.connect()
+        self.Player.server.send_immediate(
+            f"(init {self.Player.robot.name} {self.Player.world.team_name} {self.Player.world.number})"
+        )
+
+    def _safe_receive_world_update(self, retries=1):
+        last_exc = None
+        for attempt in range(retries + 1):
+            try:
+                self.Player.server.receive()
+                self.Player.world.update()
+                return
+            except (ConnectionResetError, OSError) as exc:
+                last_exc = exc
+                if attempt >= retries:
+                    raise
+                self._reconnect_server()
+        if last_exc is not None:
+            raise last_exc
+
+    def debug_log(self, message):
+        print(message)
+        try:
+            log_path = os.path.join(os.path.dirname(os.path.dirname(__file__)), "comm_debug.log")
+            with open(log_path, "a", encoding="utf-8") as f:
+                f.write(message + "\n")
+        except OSError:
+            pass
+
+    def observe(self, init=False):
+
+        """获取当前观测值"""
+        robot = self.Player.robot
+        world = self.Player.world
+
+        # Safety check: ensure data is available
+
+        # 计算目标速度
+        raw_target = self.target_position - world.global_position[:2]
+        velocity = MathOps.rotate_2d_vec(
+            raw_target,
+            -robot.global_orientation_euler[2],
+            is_rad=False
+        )
+
+        # 计算相对方向
+        rel_orientation = MathOps.vector_angle(velocity) * 0.3
+        rel_orientation = np.clip(rel_orientation, -0.25, 0.25)
+
+        velocity = np.concatenate([velocity, np.array([rel_orientation])])
+        velocity[0] = np.clip(velocity[0], -0.5, 0.5)
+        velocity[1] = np.clip(velocity[1], -0.25, 0.25)
+
+        # 关节状态
+        radian_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        radian_joint_speeds = np.deg2rad(
+            [robot.motor_speeds[motor] for motor in robot.ROBOT_MOTORS]
+        )
+
+        qpos_qvel_previous_action = np.concatenate([
+            (radian_joint_positions * self.train_sim_flip - self.joint_nominal_position) / 4.6,
+            radian_joint_speeds / 110.0 * self.train_sim_flip,
+            self.previous_action / 10.0,
+        ])
+
+        # 角速度
+        ang_vel = np.clip(np.deg2rad(robot.gyroscope) / 50.0, -1.0, 1.0)
+
+        # 投影的重力方向
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+
+        # 组合观测
+        observation = np.concatenate([
+            qpos_qvel_previous_action,
+            ang_vel,
+            velocity,
+            projected_gravity,
+        ])
+
+        observation = np.clip(observation, -10.0, 10.0)
+        return observation.astype(np.float32)
+
+    def sync(self):
+        ''' Run a single simulation step '''
+        self._safe_receive_world_update(retries=1)
+        self.Player.robot.commit_motor_targets_pd()
+        self.Player.server.send()
+        if self._target_dt > 0.0:
+            now = time.time()
+            if self._last_sync_time is None:
+                self._last_sync_time = now
+                return
+            elapsed = now - self._last_sync_time
+            remaining = self._target_dt - elapsed
+            if remaining > 0.0:
+                time.sleep(remaining)
+                now = time.time()
+            self._last_sync_time = now
+
+    def debug_joint_status(self):
+        robot = self.Player.robot
+        actual_joint_positions = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        )
+        target_joint_positions = getattr(
+            self,
+            'target_joint_positions',
+            np.zeros(len(robot.ROBOT_MOTORS), dtype=np.float32)
+        )
+        joint_error = actual_joint_positions - target_joint_positions
+        leg_slice = slice(11, None)
+
+        self.debug_log(
+            "[WalkDebug] "
+            f"step={self.step_counter} "
+            f"pos={np.round(self.Player.world.global_position, 3).tolist()} "
+            f"target_xy={np.round(self.target_position, 3).tolist()} "
+            f"target_leg={np.round(target_joint_positions[leg_slice], 3).tolist()} "
+            f"actual_leg={np.round(actual_joint_positions[leg_slice], 3).tolist()} "
+            f"err_norm={float(np.linalg.norm(joint_error)):.4f} "
+            f"fallen={self.Player.world.global_position[2] < 0.3}"
+        )
+        print(f"waist target={target_joint_positions[10]:.3f}, actual={actual_joint_positions[10]:.3f}")
+
+    def reset(self, seed=None, options=None):
+        '''
+        Reset and stabilize the robot
+        Note: for some behaviors it would be better to reduce stabilization or add noise
+        '''
+        r = self.Player.robot
+        super().reset(seed=seed)
+        if seed is not None:
+            np.random.seed(seed)
+
+        length1 = 2  # randomize target distance
+        length2 = np.random.uniform(0.6, 1)  # randomize target distance
+        length3 = np.random.uniform(0.6, 1)  # randomize target distance
+        angle2 = np.random.uniform(-30, 30)  # randomize initial orientation
+        angle3 = np.random.uniform(-30, 30)  # randomize target direction
+
+        self.step_counter = 0
+        self.waypoint_index = 0
+        self.route_completed = False
+        self.previous_action = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.last_action_for_reward = np.zeros(len(self.Player.robot.ROBOT_MOTORS))
+        self.previous_pos = np.array([0.0, 0.0])  # Initialize for first step
+        self.walk_cycle_step = 0
+
+        # 随机 beam 目标位置和朝向，增加训练多样性
+        beam_x = (random() - 0.5) * 10
+        beam_y = (random() - 0.5) * 10
+        beam_yaw = uniform(-self.reset_beam_yaw_range_deg, self.reset_beam_yaw_range_deg)
+
+        for _ in range(5):
+            self._safe_receive_world_update(retries=2)
+            self.Player.robot.commit_motor_targets_pd()
+            self.Player.server.commit_beam(pos2d=(beam_x, beam_y), rotation=beam_yaw)
+            self.Player.server.send()
+
+        # 执行 Neutral 技能直到完成，给机器人足够时间在 beam 位置稳定站立
+        finished_count = 0
+        for _ in range(50):
+            finished = self.Player.skills_manager.execute("Neutral")
+            self.sync()
+            if finished:
+                finished_count += 1
+                if finished_count >= 20:  # 假设需要连续20次完成才算成功
+                    break
+
+        if self.enable_reset_perturb and self.reset_joint_noise_rad > 0.0:
+            perturb_action = np.zeros(self.no_of_actions, dtype=np.float32)
+            # Perturb waist + lower body only (10:), keep head/arms stable.
+            perturb_action[10:] = np.random.uniform(
+                -self.reset_joint_noise_rad,
+                self.reset_joint_noise_rad,
+                size=(self.no_of_actions - 10,)
+            )
+
+            for _ in range(self.reset_perturb_steps):
+                target_joint_positions = (self.joint_nominal_position + perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+            for i in range(self.reset_recover_steps):
+                # Linearly fade perturbation to help policy start from near-neutral.
+                alpha = 1.0 - float(i + 1) / float(self.reset_recover_steps)
+                target_joint_positions = (self.joint_nominal_position + alpha * perturb_action) * self.train_sim_flip
+                for idx, target in enumerate(target_joint_positions):
+                    r.set_motor_target_position(
+                        r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=25, kd=0.6
+                    )
+                self.sync()
+
+        # memory variables
+        self.sync()  
+        self.initial_position = np.array(self.Player.world.global_position[:2])
+        self.previous_pos = self.initial_position.copy()  # Critical: set to actual position
+        self.act = np.zeros(self.no_of_actions, np.float32)
+        # Build target in the robot's current forward direction instead of fixed global +x.
+        heading_deg = float(r.global_orientation_euler[2])
+        forward_offset = MathOps.rotate_2d_vec(np.array([length1, 0.0]), heading_deg, is_rad=False)
+        point1 = self.initial_position + forward_offset
+        point2 = point1 + MathOps.rotate_2d_vec(np.array([length2, 0]), angle2, is_rad=False)
+        point3 = point2 + MathOps.rotate_2d_vec(np.array([length3, 0]), angle3, is_rad=False)
+        self.point_list = [point1]
+        self.target_position = self.point_list[self.waypoint_index]
+        self.initial_height = self.Player.world.global_position[2]
+
+        return self.observe(True), {}
+
+    def render(self, mode='human', close=False):
+        return
+
+    def compute_reward(self, previous_pos, current_pos, action):
+        height = float(self.Player.world.global_position[2])
+        robot = self.Player.robot
+
+        orientation_quat_inv = R.from_quat(robot._global_cheat_orientation).inv()
+        projected_gravity = orientation_quat_inv.apply(np.array([0.0, 0.0, -1.0]))
+        tilt_mag = float(np.linalg.norm(projected_gravity[:2]))
+        ang_vel = np.deg2rad(robot.gyroscope)
+        ang_vel_mag = float(np.linalg.norm(ang_vel))
+
+        is_fallen = height < 0.55
+        if is_fallen:
+            # remain = max(0, 800 - self.step_counter)
+            # return -8.0 - 0.01 * remain
+            return -1.0
+        
+
+
+        # # 目标方向
+        # to_target = self.target_position - current_pos
+        # dist_to_target = float(np.linalg.norm(to_target))
+        # if dist_to_target < 0.5:
+        #     return 15.0
+
+        # forward_dir = to_target / dist_to_target if dist_to_target > 0.1 else np.array([1.0, 0.0])
+        # delta_pos = current_pos - previous_pos
+        # forward_step = float(np.dot(delta_pos, forward_dir))
+        # lateral_step = float(np.linalg.norm(delta_pos - forward_dir * forward_step))
+
+        # 奖励项
+        # progress_reward = 2 * forward_step
+        # lateral_penalty = -0.1 * lateral_step
+        alive_bonus = 2.0
+
+        # action_penalty = -0.01 * float(np.linalg.norm(action))
+        smoothness_penalty = -0.01 * float(np.linalg.norm(action - self.last_action_for_reward))
+
+        posture_penalty = -0.3 * (tilt_mag)
+        ang_vel_penalty = -0.02 * ang_vel_mag
+
+        # Use simulator joint readings in training frame to shape lateral stance.
+        joint_pos = np.deg2rad(
+            [robot.motor_positions[motor] for motor in robot.ROBOT_MOTORS]
+        ) * self.train_sim_flip
+        left_hip_roll = float(joint_pos[12])
+        right_hip_roll = float(joint_pos[18])
+        left_ankle_roll = float(joint_pos[16])
+        right_ankle_roll = float(joint_pos[22])
+
+        hip_spread = left_hip_roll - right_hip_roll
+        ankle_spread = left_ankle_roll - right_ankle_roll
+        stance_metric = 0.6 * abs(hip_spread) + 0.4 * abs(ankle_spread)
+
+        # Penalize narrow stance (feet too close) and scissoring (cross-leg pattern).
+        stance_collapse_penalty = -4.0 * max(0.0, self.min_stance_rad - stance_metric)
+        cross_leg_penalty = -1.2 * max(0.0, -(hip_spread * ankle_spread))
+
+        target_height = self.initial_height
+        height_error =  height - target_height
+        height_penalty = -0.5 * abs(height_error)  # 惩罚高度偏离，系数可调
+
+        # # 在 compute_reward 开头附近，添加高度变化率计算
+        # if not hasattr(self, 'last_height'):
+        #     self.last_height = height
+        #     self.last_height_time = self.step_counter  # 可选，用于时间间隔
+        # height_rate = height - self.last_height  # 正为上升，负为下降
+        # self.last_height = height
+
+        # 惩罚高度下降（负变化率）
+        # height_down_penalty = -5.0 * max(0, -height_rate)  # 系数可调，-height_rate 为正表示下降幅度
+
+        # # 在 compute_reward 中
+        # if self.step_counter > 50:
+        #     avg_prev_action = np.mean(self.prev_action_history, axis=0)
+        #     novelty = float(np.linalg.norm(action - avg_prev_action))
+        #     exploration_bonus = 0.05 * novelty
+        # else:
+        #     exploration_bonus = 0
+
+        # self.prev_action_history[self.history_idx] = action
+        # self.history_idx = (self.history_idx + 1) % 50
+
+
+        total = (
+            # progress_reward  + 
+                 alive_bonus + 
+                 # lateral_penalty +
+                 # action_penalty +
+                smoothness_penalty +
+                 posture_penalty
+                 + ang_vel_penalty
+                 + height_penalty
+                 + stance_collapse_penalty
+                 + cross_leg_penalty
+                #  + exploration_bonus
+                # + height_down_penalty
+                 )
+        if time.time() - self.start_time >= 600:
+            self.start_time = time.time()
+            print(
+                #   f"progress_reward:{progress_reward:.4f}",
+                #   f"lateral_penalty:{lateral_penalty:.4f}",
+                #   f"action_penalty:{action_penalty:.4f}"s, 
+                  f"height_penalty:{height_penalty:.4f}",
+                  f"smoothness_penalty:{smoothness_penalty:.4f},",
+                  f"posture_penalty:{posture_penalty:.4f}",
+                  f"stance_collapse_penalty:{stance_collapse_penalty:.4f}",
+                  f"cross_leg_penalty:{cross_leg_penalty:.4f}",
+                #   f"ang_vel_penalty:{ang_vel_penalty:.4f}",
+                #   f"height_down_penalty:{height_down_penalty:.4f}",
+                #   f"exploration_bonus:{exploration_bonus:.4f}"
+            )
+
+        return total
+
+
+
+    def step(self, action):
+
+        r = self.Player.robot
+        self.previous_action = action
+
+        self.target_joint_positions = (
+                # self.joint_nominal_position +
+                 self.scaling_factor * action
+        )
+        self.target_joint_positions *= self.train_sim_flip
+
+        for idx, target in enumerate(self.target_joint_positions):
+            r.set_motor_target_position(
+                r.ROBOT_MOTORS[idx], target * 180 / math.pi, kp=40, kd=1.0
+            )
+
+        self.previous_action = action
+
+        self.sync()  # run simulation step
+        self.step_counter += 1
+
+        if self.enable_debug_joint_status and self.step_counter % self.debug_every_n_steps == 0:
+            self.debug_joint_status()
+
+        current_pos = np.array(self.Player.world.global_position[:2], dtype=np.float32)
+
+        # Compute reward based on movement from previous step
+        reward = self.compute_reward(self.previous_pos, current_pos, action)
+
+        # Update previous position
+        self.previous_pos = current_pos.copy()
+        self.last_action_for_reward = action.copy()
+
+        # Fall detection and penalty
+        is_fallen = self.Player.world.global_position[2] < 0.55
+
+        # terminal state: the robot is falling or timeout
+        terminated = is_fallen or self.step_counter > 800 or self.route_completed
+        truncated = False
+
+        return self.observe(), reward, terminated, truncated, {}
+
+
+class Train(Train_Base):
+    def __init__(self, script) -> None:
+        super().__init__(script)
+
+    def train(self, args):
+
+        # --------------------------------------- Learning parameters
+        n_envs = int(os.environ.get("GYM_CPU_N_ENVS", "20"))
+        if n_envs < 1:
+            raise ValueError("GYM_CPU_N_ENVS must be >= 1")
+        server_warmup_sec = float(os.environ.get("GYM_CPU_SERVER_WARMUP_SEC", "3.0"))
+        n_steps_per_env = int(os.environ.get("GYM_CPU_TRAIN_STEPS_PER_ENV", "256"))  # RolloutBuffer is of size (n_steps_per_env * n_envs)
+        minibatch_size = int(os.environ.get("GYM_CPU_TRAIN_BATCH_SIZE", "512"))  # should be a factor of (n_steps_per_env * n_envs)
+        total_steps = 30000000
+        learning_rate = float(os.environ.get("GYM_CPU_TRAIN_LR", "3e-4"))
+        folder_name = f'Walk_R{self.robot_type}'
+        model_path = f'./scripts/gyms/logs/{folder_name}/'
+
+        print(f"Model path: {model_path}")
+        print(f"Using {n_envs} parallel environments")
+
+        # --------------------------------------- Run algorithm
+        def init_env(i_env, monitor=False):
+            def thunk():
+                env = WalkEnv(self.ip, self.server_p + i_env)
+                if monitor:
+                    env = Monitor(env)
+                return env
+
+            return thunk
+
+        server_log_dir = os.path.join(model_path, "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        servers = Train_Server(self.server_p, self.monitor_p_1000, n_envs + 1, no_render=True, no_realtime=True)  # include 1 extra server for testing
+
+        # Wait for servers to start
+        print(f"Starting {n_envs + 1} rcssservermj servers...")
+        if server_warmup_sec > 0:
+            print(f"Waiting {server_warmup_sec:.1f}s for server warmup...")
+            sleep(server_warmup_sec)
+        print("Servers started, creating environments...")
+
+        env = SubprocVecEnv([init_env(i, monitor=True) for i in range(n_envs)])
+        # Use single-process eval env to avoid extra subprocess fragility during callback evaluation.
+        eval_env = DummyVecEnv([init_env(n_envs, monitor=True)])
+
+        try:
+            # Custom policy network architecture
+            policy_kwargs = dict(
+                net_arch=dict(
+                    pi=[512, 256, 128],  # Policy network: 3 layers
+                    vf=[512, 256, 128]  # Value network: 3 layers
+                ),
+                activation_fn=__import__('torch.nn', fromlist=['ELU']).ELU,
+            )
+
+            if "model_file" in args:  # retrain
+                model = PPO.load(args["model_file"], env=env, device="cpu", n_envs=n_envs, n_steps=n_steps_per_env,
+                                 batch_size=minibatch_size, learning_rate=learning_rate)
+            else:  # train new model
+                model = PPO(
+                    "MlpPolicy",
+                    env=env,
+                    verbose=1,
+                    n_steps=n_steps_per_env,
+                    batch_size=minibatch_size,
+                    learning_rate=learning_rate,
+                    device="cpu",
+                    policy_kwargs=policy_kwargs,
+                    ent_coef=float(os.environ.get("GYM_CPU_TRAIN_ENT_COEF", "0.05")),  # Entropy coefficient for exploration
+                    clip_range=float(os.environ.get("GYM_CPU_TRAIN_CLIP_RANGE", "0.2")),  # PPO clipping parameter
+                    gae_lambda=0.95,  # GAE lambda
+                    gamma=float(os.environ.get("GYM_CPU_TRAIN_GAMMA", "0.95")), # Discount factor
+                    # target_kl=0.03,
+                    n_epochs=int(os.environ.get("GYM_CPU_TRAIN_EPOCHS", "5")),
+                    tensorboard_log=f"./scripts/gyms/logs/{folder_name}/tensorboard/"
+                )
+
+            model_path = self.learn_model(model, total_steps, model_path, eval_env=eval_env,
+                                          eval_freq=n_steps_per_env * 20, save_freq=n_steps_per_env * 20, eval_eps=100,
+                                          backup_env_file=__file__)
+        except KeyboardInterrupt:
+            sleep(1)  # wait for child processes
+            print("\nctrl+c pressed, aborting...\n")
+            servers.kill()
+            return
+
+        env.close()
+        eval_env.close()
+        servers.kill()
+
+    def test(self, args):
+
+        # Uses different server and monitor ports
+        server_log_dir = os.path.join(args["folder_dir"], "server_logs")
+        os.makedirs(server_log_dir, exist_ok=True)
+        test_no_render = os.environ.get("GYM_CPU_TEST_NO_RENDER", "0") == "1"
+        test_no_realtime = os.environ.get("GYM_CPU_TEST_NO_REALTIME", "0") == "1"
+
+        server = Train_Server(
+            self.server_p - 1,
+            self.monitor_p,
+            1,
+            no_render=test_no_render,
+            no_realtime=test_no_realtime,
+        )
+        env = WalkEnv(self.ip, self.server_p - 1)
+        model = PPO.load(args["model_file"], env=env)
+
+        try:
+            self.export_model(args["model_file"], args["model_file"] + ".pkl",
+                              False)  # Export to pkl to create custom behavior
+            self.test_model(model, env, log_path=args["folder_dir"], model_path=args["folder_dir"])
+        except KeyboardInterrupt:
+            print()
+
+        env.close()
+        server.kill()
+
+
+if __name__ == "__main__":
+    from types import SimpleNamespace
+
+    # 创建默认参数
+    script_args = SimpleNamespace(
+        args=SimpleNamespace(
+            i='127.0.0.1',  # Server IP
+            p=3100,  # Server port
+            m=3200,  # Monitor port
+            r=0,  # Robot type
+            t='Gym',  # Team name
+            u=1  # Uniform number
+        )
+    )
+
+    trainer = Train(script_args)
+
+    run_mode = os.environ.get("GYM_CPU_MODE", "train").strip().lower()
+
+    if run_mode == "test":
+        test_model_file = os.environ.get("GYM_CPU_TEST_MODEL", "scripts/gyms/logs/Walk_R0_004/best_model.zip")
+        test_folder = os.environ.get("GYM_CPU_TEST_FOLDER", "scripts/gyms/logs/Walk_R0_004/")
+        trainer.test({"model_file": test_model_file, "folder_dir": test_folder})
+    else:
+        retrain_model = os.environ.get("GYM_CPU_TRAIN_MODEL", "").strip()
+        if retrain_model:
+            trainer.train({"model_file": retrain_model})
+        else:
+            trainer.train({})

train.sh

@@ -0,0 +1,136 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+# ------------------------------
+# 资源限制配置（cgroup v2 + systemd-run）
+# ------------------------------
+# 说明：
+# 1) 这个脚本会把训练进程放进一个临时的 systemd scope 中，并施加 CPU/内存上限。
+# 2) 仅限制“本次训练进程”，不会永久改系统配置。
+# 3) 下面变量都支持“环境变量覆盖”，即你可以在命令前临时指定。
+#
+# CPU 核数基准（默认 20）：
+# 例如你的机器按 20 核预算来算，可保持默认。
+CORES="${CORES:-20}"
+# CPU 占用百分比（默认 100）：
+# 最终会与 CORES 相乘得到 CPUQuota。
+# 例：CORES=20, UTIL_PERCENT=100 -> CPUQuota=2000%（约 20 核等效）
+UTIL_PERCENT="${UTIL_PERCENT:-100}"
+CPU_QUOTA="$((CORES * UTIL_PERCENT))%"
+
+# 内存上限（默认关闭）：
+# 设为具体值（如 24G/28G）可限制训练最多占用内存；
+# 设为 0/none/off/infinity 表示不设置 cgroup 内存上限。
+MEMORY_MAX="${MEMORY_MAX:-0}"
+
+# ------------------------------
+# 训练运行参数（由 scripts/gyms/Walk.py 读取）
+# ------------------------------
+# 运行模式：train 或 test
+GYM_CPU_MODE="${GYM_CPU_MODE:-train}"
+
+# 并行环境数量：越大通常吞吐越高，但也更容易触发 OOM 或连接不稳定。
+# 默认使用更稳妥的 12，确认稳定后再升到 16/20。
+GYM_CPU_N_ENVS="${GYM_CPU_N_ENVS:-20}"
+# 服务器预热时间（秒）：
+# 在批量拉起 rcssserver 后等待一段时间，再创建 SubprocVecEnv，
+# 可降低 ConnectionReset/EOFError 概率。
+GYM_CPU_SERVER_WARMUP_SEC="${GYM_CPU_SERVER_WARMUP_SEC:-10}"
+
+# 训练专用参数
+GYM_CPU_TRAIN_STEPS_PER_ENV="${GYM_CPU_TRAIN_STEPS_PER_ENV:-256}"
+GYM_CPU_TRAIN_BATCH_SIZE="${GYM_CPU_TRAIN_BATCH_SIZE:-512}"
+GYM_CPU_TRAIN_LR="${GYM_CPU_TRAIN_LR:-1e-4}"
+GYM_CPU_TRAIN_ENT_COEF="${GYM_CPU_TRAIN_ENT_COEF:-0.03}"
+GYM_CPU_TRAIN_CLIP_RANGE="${GYM_CPU_TRAIN_CLIP_RANGE:-0.13}"
+GYM_CPU_TRAIN_GAMMA="${GYM_CPU_TRAIN_GAMMA:-0.95}"
+GYM_CPU_TRAIN_EPOCHS="${GYM_CPU_TRAIN_EPOCHS:-5}"
+GYM_CPU_TRAIN_MODEL="${GYM_CPU_TRAIN_MODEL:-}"
+
+# 测试专用参数
+GYM_CPU_TEST_MODEL="${GYM_CPU_TEST_MODEL:-scripts/gyms/logs/Walk_R0_004/best_model.zip}"
+GYM_CPU_TEST_FOLDER="${GYM_CPU_TEST_FOLDER:-scripts/gyms/logs/Walk_R0_004/}"
+# 测试默认实时且显示画面：默认均为 0
+# 设为 1 表示关闭对应能力
+GYM_CPU_TEST_NO_RENDER="${GYM_CPU_TEST_NO_RENDER:-0}"
+GYM_CPU_TEST_NO_REALTIME="${GYM_CPU_TEST_NO_REALTIME:-0}"
+
+# Python 解释器选择策略：
+# 1) 优先使用你手动传入的 PYTHON_BIN
+# 2) 其次用当前激活 conda 环境（CONDA_PREFIX/bin/python）
+# 3) 再回退到默认 mujoco 环境路径
+# 4) 最后尝试系统 python / python3
+DEFAULT_PYTHON="/home/solren/Downloads/Anaconda/envs/mujoco/bin/python"
+CONDA_PYTHON="${CONDA_PREFIX:-}/bin/python"
+
+# 安全保护：不要用 sudo 运行。
+# 原因：sudo 可能导致 conda 环境与用户会话环境不一致，
+# 会引发 python 路径丢失、systemd --user 会话不可见等问题。
+if [[ "${EUID}" -eq 0 ]]; then
+	echo "Do not run this script with sudo; run as your normal user in conda env 'mujoco'."
+	exit 1
+fi
+
+# 解析最终使用的 Python 可执行文件。
+if [[ -n "${PYTHON_BIN:-}" ]]; then
+	PYTHON_EXEC="${PYTHON_BIN}"
+elif [[ -n "${CONDA_PREFIX:-}" && -x "${CONDA_PYTHON}" ]]; then
+	PYTHON_EXEC="${CONDA_PYTHON}"
+elif [[ -x "${DEFAULT_PYTHON}" ]]; then
+	PYTHON_EXEC="${DEFAULT_PYTHON}"
+elif command -v python >/dev/null 2>&1; then
+	PYTHON_EXEC="$(command -v python)"
+elif command -v python3 >/dev/null 2>&1; then
+	PYTHON_EXEC="$(command -v python3)"
+else
+	echo "No Python executable found. Set PYTHON_BIN=/abs/path/to/python and retry."
+	exit 1
+fi
+
+# 脚本所在目录（绝对路径），便于后续定位模块/相对路径。
+SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
+
+# 打印当前生效配置，方便排障和复现实验。
+echo "Starting training with limits: CPU=${CPU_QUOTA}, Memory=${MEMORY_MAX}"
+echo "Mode: ${GYM_CPU_MODE}"
+echo "Runtime knobs: GYM_CPU_N_ENVS=${GYM_CPU_N_ENVS}, GYM_CPU_SERVER_WARMUP_SEC=${GYM_CPU_SERVER_WARMUP_SEC}"
+echo "Using Python: ${PYTHON_EXEC}"
+if [[ -n "${CONDA_DEFAULT_ENV:-}" ]]; then
+	echo "Detected conda env: ${CONDA_DEFAULT_ENV}"
+fi
+
+SYSTEMD_PROPS=("-p" "CPUQuota=${CPU_QUOTA}")
+case "${MEMORY_MAX,,}" in
+	0|none|off|infinity)
+		echo "MemoryMax is disabled for this run (no cgroup memory cap)."
+		;;
+	*)
+		SYSTEMD_PROPS+=("-p" "MemoryMax=${MEMORY_MAX}")
+		;;
+esac
+
+# 使用 systemd-run --user --scope 启动“受限资源”的训练进程：
+# - CPUQuota: 总 CPU 配额
+# - MemoryMax: 最大内存
+# - env ...  : 显式传递训练参数到 Python 进程
+# - python -m scripts.gyms.Walk: 以模块方式启动训练入口
+systemd-run --user --scope \
+	"${SYSTEMD_PROPS[@]}" \
+	env \
+	GYM_CPU_MODE="${GYM_CPU_MODE}" \
+	GYM_CPU_N_ENVS="${GYM_CPU_N_ENVS}" \
+	GYM_CPU_SERVER_WARMUP_SEC="${GYM_CPU_SERVER_WARMUP_SEC}" \
+	GYM_CPU_TRAIN_STEPS_PER_ENV="${GYM_CPU_TRAIN_STEPS_PER_ENV}" \
+	GYM_CPU_TRAIN_BATCH_SIZE="${GYM_CPU_TRAIN_BATCH_SIZE}" \
+	GYM_CPU_TRAIN_LR="${GYM_CPU_TRAIN_LR}" \
+	GYM_CPU_TRAIN_ENT_COEF="${GYM_CPU_TRAIN_ENT_COEF}" \
+	GYM_CPU_TRAIN_CLIP_RANGE="${GYM_CPU_TRAIN_CLIP_RANGE}" \
+	GYM_CPU_TRAIN_GAMMA="${GYM_CPU_TRAIN_GAMMA}" \
+	GYM_CPU_TRAIN_EPOCHS="${GYM_CPU_TRAIN_EPOCHS}" \
+	GYM_CPU_TRAIN_MODEL="${GYM_CPU_TRAIN_MODEL}" \
+	GYM_CPU_TEST_MODEL="${GYM_CPU_TEST_MODEL}" \
+	GYM_CPU_TEST_FOLDER="${GYM_CPU_TEST_FOLDER}" \
+	GYM_CPU_TEST_NO_RENDER="${GYM_CPU_TEST_NO_RENDER}" \
+	GYM_CPU_TEST_NO_REALTIME="${GYM_CPU_TEST_NO_REALTIME}" \
+	"${PYTHON_EXEC}" "-m" "scripts.gyms.Walk"
+

world/world.py

@@ -47,6 +47,7 @@ class World:
         self.their_team_players: list[OtherRobot] = [OtherRobot(is_teammate=False) for _ in
                                                      range(self.MAX_PLAYERS_PER_TEAM)]
         self.field: Field = self.__initialize_field(field_name=field_name)
+        self.WORLD_STEPTIME: float = 0.005  # Time step of the world in seconds
 
     def update(self) -> None:
         """