new mimic

mimic
Her-darling 4 weeks ago
parent 6c9467f21e
commit 7d49579686

@ -8,13 +8,14 @@ import pickle
class Dribble(): class Dribble():
def __init__(self, base_agent : Base_Agent) -> None: def __init__(self, base_agent: Base_Agent) -> None:
self.behavior = base_agent.behavior self.behavior = base_agent.behavior
self.path_manager = base_agent.path_manager self.path_manager = base_agent.path_manager
self.world = base_agent.world self.world = base_agent.world
self.description = "RL dribble" self.description = "RL dribble"
self.auto_head = True self.auto_head = True
self.env = Env(base_agent, 0.9 if self.world.robot.type == 3 else 1.2) self.env = Env(base_agent, 0.9 if self.world.robot.type == 3 else 1.2)
self.expert_data = []
with open(M.get_active_directory([ with open(M.get_active_directory([
"/behaviors/custom/Dribble/dribble_long_R1_00_178M.pkl", "/behaviors/custom/Dribble/dribble_long_R1_00_178M.pkl",
"/behaviors/custom/Dribble/dribble_long_R1_00_178M.pkl", "/behaviors/custom/Dribble/dribble_long_R1_00_178M.pkl",
@ -177,7 +178,8 @@ class Dribble():
#------------------------ 2. Execute behavior #------------------------ 2. Execute behavior
obs = self.env.observe(reset_dribble) obs = self.env.observe(reset_dribble)
action = run_mlp(obs, self.model) action = run_mlp(obs, self.model)
self.expert_data.append((obs, action))
self.env.execute(action) self.env.execute(action)
# wind down dribbling, and then reset phase # wind down dribbling, and then reset phase
@ -195,6 +197,8 @@ class Dribble():
#------------------------ 3. Reset behavior #------------------------ 3. Reset behavior
self.phase += 1 self.phase += 1
if self.phase >= WIND_DOWN_STEPS - 5: if self.phase >= WIND_DOWN_STEPS - 5:
with open('dribble_expert_data.pkl', 'wb') as f:
pickle.dump(self.expert_data, f) # 保存观测值到 pkl 文件
self.phase = 0 self.phase = 0
return True return True

@ -0,0 +1,13 @@
import pickle
# 加载并查看 pkl 文件
def load_and_display_observations(file_path):
with open(file_path, 'rb') as f:
observations = pickle.load(f)
# 打印观测值
for i, obs in enumerate(observations):
print(f"Observation {i + 1}: {obs}")
print(type(observations))
# 使用函数查看指定的 pkl 文件
load_and_display_observations('scripts/gyms/dribble_expert_data.pkl')

@ -1,7 +1,6 @@
import math import math
import random import random
import pickle
from math_ops.Neural_Network import run_mlp
from agent.Base_Agent import Base_Agent as Agent from agent.Base_Agent import Base_Agent as Agent
from behaviors.custom.Step.Step import Step from behaviors.custom.Step.Step import Step
from world.commons.Draw import Draw from world.commons.Draw import Draw
@ -12,10 +11,8 @@ from scripts.commons.Train_Base import Train_Base
from time import sleep from time import sleep
import os, gym import os, gym
import numpy as np import numpy as np
from math_ops.Math_Ops import Math_Ops as U
from math_ops.Math_Ops import Math_Ops as M from math_ops.Math_Ops import Math_Ops as M
from behaviors.custom.Step.Step_Generator import Step_Generator from behaviors.custom.Step.Step_Generator import Step_Generator
import torch
''' '''
Objective: Objective:
@ -260,16 +257,19 @@ class dribble(gym.Env):
Draw.clear_all() Draw.clear_all()
self.player.terminate() self.player.terminate()
def execute(self, action): def step(self, action):
r = (self.
player.world.robot)
# Actions: w = self.player.world
# 0,1,2 left ankle pos d = w.draw
# 3,4,5 right ankle pos if w.ball_abs_pos[1] > 0: #
# 6,7,8 left foot rotation dribble_target = (15, 5)
# 9,10,11 right foot rotation else:
# 12,13 left/right arm pitch dribble_target = (15, -5)
# 14,15 left/right arm roll
self.dribble_rel_orientation = self.path_manager.get_dribble_path(optional_2d_target=dribble_target)[1]
# exponential moving average # exponential moving average
self.act = 0.85 * self.act + 0.15 * action * 0.7 * 0.95 * self.dribble_speed self.act = 0.85 * self.act + 0.15 * action * 0.7 * 0.95 * self.dribble_speed
@ -293,49 +293,6 @@ class dribble(gym.Env):
arms = np.copy(self.DEFAULT_ARMS) # default arms pose arms = np.copy(self.DEFAULT_ARMS) # default arms pose
arms[0:4] += a[12:16] * 4 # arms pitch+roll arms[0:4] += a[12:16] * 4 # arms pitch+roll
return l_ankle_pos, r_ankle_pos, l_foot_rot, r_foot_rot, arms
def loss(self, obs, action_p, action_r):
r = self.player.world.robot
with open(M.get_active_directory([
"/behaviors/custom/Dribble/dribble_long_R1_00_178M.pkl",
"/behaviors/custom/Dribble/dribble_long_R1_00_178M.pkl",
"/behaviors/custom/Dribble/dribble_long_R1_00_178M.pkl",
"/behaviors/custom/Dribble/dribble_long_R1_00_178M.pkl",
"/behaviors/custom/Dribble/dribble_long_R1_00_178M.pkl"
][r.type]), 'rb') as f:
model = pickle.load(f)
act = run_mlp(obs, model)
a_l_ankle_pos, a_r_ankle_pos, a_l_foot_rot, a_r_foot_rot, a_arms = self.execute(act)
act_p = np.concatenate((a_l_ankle_pos, a_r_ankle_pos))
act_r = np.concatenate((a_l_foot_rot, a_r_foot_rot, a_arms))
action_p_tensor = torch.from_numpy(action_p)
action_r_tensor = torch.from_numpy(action_r)
act_p_tensor = torch.from_numpy(act_p)
act_r_tensor = torch.from_numpy(act_r)
loss_p = torch.exp(-40*torch.norm(action_p_tensor - act_p_tensor, p=2))
loss_r = torch.exp(-0.1*torch.norm(action_r_tensor - act_r_tensor, p=2))
loss = loss_p + loss_r
return loss
def step(self, action):
r = (self.
player.world.robot)
w = self.player.world
d = w.draw
if w.ball_abs_pos[1] > 0: #
dribble_target = (15, 5)
else:
dribble_target = (15, -5)
self.dribble_rel_orientation = self.path_manager.get_dribble_path(optional_2d_target=dribble_target)[1]
l_ankle_pos, r_ankle_pos, l_foot_rot, r_foot_rot, arms = self.execute(action)
action_p = np.concatenate((l_ankle_pos, r_ankle_pos))
action_r = np.concatenate((l_foot_rot, r_foot_rot, arms))
# Set target positions # Set target positions
self.execute_ik(l_ankle_pos, l_foot_rot, r_ankle_pos, r_foot_rot) # legs self.execute_ik(l_ankle_pos, l_foot_rot, r_ankle_pos, r_foot_rot) # legs
r.set_joints_target_position_direct(slice(14, 22), arms, harmonize=False) # arms r.set_joints_target_position_direct(slice(14, 22), arms, harmonize=False) # arms
@ -351,11 +308,8 @@ class dribble(gym.Env):
np.linalg.norm(unit_vector) * np.linalg.norm(w.ball_cheat_abs_vel[:2])) np.linalg.norm(unit_vector) * np.linalg.norm(w.ball_cheat_abs_vel[:2]))
else: else:
cos_theta = 0 cos_theta = 0
loss = self.loss(obs, action_p, action_r)
# 计算奖励 # 计算奖励
reward = np.linalg.norm(w.ball_cheat_abs_vel) * cos_theta + loss reward = np.linalg.norm(w.ball_cheat_abs_vel) * cos_theta
if self.ball_dist_hip_center_2d < 0.115: if self.ball_dist_hip_center_2d < 0.115:
reward = 0 reward = 0
@ -377,7 +331,7 @@ class Train(Train_Base):
def train(self, args): def train(self, args):
# --------------------------------------- Learning parameters # --------------------------------------- Learning parameters
n_envs = min(12, os.cpu_count()) n_envs = min(1, os.cpu_count())
n_steps_per_env = 1024 # RolloutBuffer is of size (n_steps_per_env * n_envs) n_steps_per_env = 1024 # RolloutBuffer is of size (n_steps_per_env * n_envs)
minibatch_size = 64 # should be a factor of (n_steps_per_env * n_envs) minibatch_size = 64 # should be a factor of (n_steps_per_env * n_envs)
total_steps = 50000000 total_steps = 50000000

@ -0,0 +1,439 @@
import math
import pickle
import random
from imitation.rewards.reward_nets import BasicRewardNet
from stable_baselines3.common.callbacks import EvalCallback
from agent.Base_Agent import Base_Agent as Agent
from behaviors.custom.Step.Step import Step
from world.commons.Draw import Draw
from stable_baselines3 import PPO
from stable_baselines3.common.vec_env import SubprocVecEnv
from scripts.commons.Server import Server
from scripts.commons.Train_Base import Train_Base
from time import sleep
import os, gym
import numpy as np
from math_ops.Math_Ops import Math_Ops as M
from behaviors.custom.Step.Step_Generator import Step_Generator
from imitation.data import rollout
from imitation.algorithms.adversarial.gail import GAIL
'''
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 dribble(gym.Env):
def __init__(self, ip, server_p, monitor_p, r_type, enable_draw) -> None:
self.abs_ori = 75
self.ball_dist_hip_center_2d = 0
self.ball_dist_hip_center = None
self.internal_rel_orientation = None
self.dribble_speed = 1
self.gym_last_internal_abs_ori = None
self.internal_target_vel = None
self.Gen_player_pos = None
self.internal_target = None
self.values_l = None
self.values_r = None
self.reset_time = None
self.behavior = None
self.robot_type = r_type
self.kick_ori = 0
self.terminal = False
# Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
self.player = Agent(ip, server_p, monitor_p, 1, self.robot_type, "Gym", True, True)
self.step_counter = 0 # to limit episode size
self.ik = self.player.inv_kinematics
self.dribble_rel_orientation = 0 # relative to imu_torso_orientation (in degrees)
# Step behavior defaults
self.STEP_DUR = 8
self.STEP_Z_SPAN = 0.02
self.STEP_Z_MAX = 0.70
nao_specs = self.ik.NAO_SPECS
self.leg_length = nao_specs[1] + nao_specs[3] # upper leg height + lower leg height
feet_y_dev = nao_specs[0] * 1.12 # wider step
sample_time = self.player.world.robot.STEPTIME
max_ankle_z = nao_specs[5]
self.step_generator = Step_Generator(feet_y_dev, sample_time, max_ankle_z)
self.DEFAULT_ARMS = np.array([-90, -90, 8, 8, 90, 90, 70, 70], np.float32)
self.act = np.zeros(16, np.float32) # memory variable
self.path_manager = self.player.path_manager
# State space
obs_size = 76
self.obs = np.zeros(obs_size, np.float32)
self.observation_space = gym.spaces.Box(low=np.full(obs_size, -np.inf, np.float32),
high=np.full(obs_size, np.inf, np.float32), dtype=np.float32)
self.ball_x_center = 0.21
self.ball_y_center = -0.045
# Action space
MAX = np.finfo(np.float32).max
self.no_of_actions = act_size = 16
self.action_space = gym.spaces.Box(low=np.full(act_size, -MAX, np.float32),
high=np.full(act_size, MAX, np.float32), dtype=np.float32)
# Place ball far away to keep landmarks in FoV (head follows ball while using Step behavior)
self.player.scom.unofficial_move_ball((14, 0, 0.042))
self.player.scom.unofficial_set_play_mode("PlayOn")
self.player.scom.unofficial_move_ball((0, 0, 0))
def observe(self, init=False, virtual_ball=False):
r = self.player.world.robot
w = self.player.world
if init: # reset variables
self.step_counter = 0
self.act = np.zeros(16, np.float32) # memory variable
# index observation naive normalization
self.obs[0] = min(self.step_counter, 12 * 8) / 100 # simple counter: 0,1,2,3...
self.obs[1] = r.loc_head_z * 3 # z coordinate (torso)
self.obs[2] = r.loc_head_z_vel / 2 # z velocity (torso)
self.obs[3] = r.imu_torso_roll / 15 # absolute torso roll in deg
self.obs[4] = r.imu_torso_pitch / 15 # absolute torso pitch in deg
self.obs[5:8] = r.gyro / 100 # gyroscope
self.obs[8:11] = r.acc / 10 # accelerometer
self.obs[11:17] = r.frp.get('lf', np.zeros(6)) * (10, 10, 10, 0.01, 0.01,
0.01) # left foot: relative point of origin (p) and force vector (f) -> (px,py,pz,fx,fy,fz)*
self.obs[17:23] = r.frp.get('rf', np.zeros(6)) * (10, 10, 10, 0.01, 0.01,
0.01) # right foot: relative point of origin (p) and force vector (f) -> (px,py,pz,fx,fy,fz)*
# *if foot is not touching the ground, then (px=0,py=0,pz=0,fx=0,fy=0,fz=0)
self.obs[23:43] = r.joints_position[2:22] / 100 # position of all joints except head & toes (for robot type 4)
self.obs[43:63] = r.joints_speed[2:22] / 6.1395 # speed of all joints except head & toes (for robot type 4)
if init: # the walking parameters refer to the last parameters in effect (after a reset, they are pointless)
self.obs[63] = 1 # step progress
self.obs[64] = 1 # 1 if left leg is active
self.obs[65] = 0 # 1 if right leg is active
self.obs[66] = 0
else:
self.obs[63] = self.step_generator.external_progress # step progress
self.obs[64] = float(self.step_generator.state_is_left_active) # 1 if left leg is active
self.obs[65] = float(not self.step_generator.state_is_left_active) # 1 if right leg is active
self.obs[66] = math.sin(self.step_generator.state_current_ts / self.step_generator.ts_per_step * math.pi)
# Ball
ball_rel_hip_center = self.ik.torso_to_hip_transform(w.ball_rel_torso_cart_pos)
self.ball_dist_hip_center = np.linalg.norm(ball_rel_hip_center)
ball_rel_torso_xy = w.ball_rel_torso_cart_pos[:2] # 取X和Y分量
self.ball_dist_hip_center_2d = np.linalg.norm(ball_rel_torso_xy)
if init:
self.obs[67:70] = (0, 0, 0) # Initial velocity is 0
elif w.ball_is_visible:
self.obs[67:70] = (ball_rel_hip_center - self.obs[
70:73]) * 10 # Ball velocity, relative to ankle's midpoint
self.obs[70:73] = ball_rel_hip_center # Ball position, relative to hip
self.obs[73] = self.ball_dist_hip_center * 2
if virtual_ball: # simulate the ball between the robot's feet
self.obs[67:74] = (0, 0, 0, 0.05, 0, -0.175, 0.36)
'''
Create internal target with a smoother variation
'''
MAX_ROTATION_DIFF = 20 # max difference (degrees) per visual step
MAX_ROTATION_DIST = 80
if init:
self.internal_rel_orientation = 0
self.internal_target_vel = 0
self.gym_last_internal_abs_ori = r.imu_torso_orientation # for training purposes (reward)
# ---------------------------------------------------------------- compute internal target
if w.vision_is_up_to_date:
previous_internal_rel_orientation = np.copy(self.internal_rel_orientation)
internal_ori_diff = np.clip(M.normalize_deg(self.dribble_rel_orientation - self.internal_rel_orientation),
-MAX_ROTATION_DIFF, MAX_ROTATION_DIFF)
self.internal_rel_orientation = np.clip(M.normalize_deg(self.internal_rel_orientation + internal_ori_diff),
-MAX_ROTATION_DIST, MAX_ROTATION_DIST)
# Observations
self.internal_target_vel = self.internal_rel_orientation - previous_internal_rel_orientation
self.gym_last_internal_abs_ori = self.internal_rel_orientation + r.imu_torso_orientation
# ----------------------------------------------------------------- observations
self.obs[74] = self.internal_rel_orientation / MAX_ROTATION_DIST
self.obs[75] = self.internal_target_vel / MAX_ROTATION_DIFF
return self.obs
def execute_ik(self, l_pos, l_rot, r_pos, r_rot):
r = self.player.world.robot
# Apply IK to each leg + Set joint targets
# Left leg
indices, self.values_l, error_codes = self.ik.leg(l_pos, l_rot, True, dynamic_pose=False)
r.set_joints_target_position_direct(indices, self.values_l, harmonize=False)
# Right leg
indices, self.values_r, error_codes = self.ik.leg(r_pos, r_rot, False, dynamic_pose=False)
r.set_joints_target_position_direct(indices, self.values_r, harmonize=False)
def sync(self):
''' Run a single simulation step '''
r = self.player.world.robot
self.player.scom.commit_and_send(r.get_command())
self.player.scom.receive()
def reset(self):
# print("reset")
'''
Reset and stabilize the robot
Note: for some behaviors it would be better to reduce stabilization or add noise
'''
self.abs_ori = 45
self.player.scom.unofficial_set_play_mode("PlayOn")
Gen_ball_pos = [- 9, 0, 0]
self.Gen_player_pos = (Gen_ball_pos[0] - 1.5, Gen_ball_pos[1], 0.5)
self.player.scom.unofficial_move_ball((Gen_ball_pos[0], Gen_ball_pos[1], Gen_ball_pos[2]))
self.step_counter = 0
self.behavior = self.player.behavior
r = self.player.world.robot
w = self.player.world
t = w.time_local_ms
self.reset_time = t
self.b_rel = w.ball_rel_torso_cart_pos[:2]
self.path_manager = self.player.path_manager
for _ in range(25):
self.player.scom.unofficial_beam(self.Gen_player_pos, 0) # beam player continuously (floating above ground)
self.player.behavior.execute("Zero_Bent_Knees")
self.sync()
# beam player to ground
self.player.scom.unofficial_beam(self.Gen_player_pos, 0)
r.joints_target_speed[
0] = 0.01 # move head to trigger physics update (rcssserver3d bug when no joint is moving)
self.sync()
# stabilize on ground
for _ in range(7):
self.player.behavior.execute("Zero_Bent_Knees")
self.sync()
# walk to ball
while True:
self.b_rel = w.ball_rel_torso_cart_pos[:2]
if self.player.behavior.is_ready("Get_Up"):
self.player.behavior.execute_to_completion("Get_Up")
if 0.26 > self.b_rel[0] > 0.18 and abs(self.b_rel[1]) < 0.04 and w.ball_is_visible:
break
else:
if self.player.behavior.is_ready("Get_Up"):
self.player.behavior.execute_to_completion("Get_Up")
reset_walk = self.behavior.previous_behavior != "Walk" # reset walk if it wasn't the previous behavior
rel_target = self.b_rel + (-0.23, 0) # relative target is a circle (center: ball, radius:0.23m)
rel_ori = M.vector_angle(self.b_rel) # relative orientation of ball, NOT the target!
dist = max(0.08, np.linalg.norm(rel_target) * 0.7) # slow approach
self.behavior.execute_sub_behavior("Walk", reset_walk, rel_target, False, rel_ori, False,
dist) # target, is_target_abs, ori, is_ori_abs, distance
self.sync()
self.act = np.zeros(self.no_of_actions, np.float32)
return self.observe(True)
def render(self, mode='human', close=False):
return
def close(self):
Draw.clear_all()
self.player.terminate()
def step(self, action):
r = (self.
player.world.robot)
w = self.player.world
d = w.draw
if w.ball_abs_pos[1] > 0: #
dribble_target = (15, 5)
else:
dribble_target = (15, -5)
self.dribble_rel_orientation = self.path_manager.get_dribble_path(optional_2d_target=dribble_target)[1]
# exponential moving average
self.act = 0.85 * self.act + 0.15 * action * 0.7 * 0.95 * self.dribble_speed
# execute Step behavior to extract the target positions of each leg (we will override these targets)
lfy, lfz, rfy, rfz = self.step_generator.get_target_positions(self.step_counter == 0, self.STEP_DUR,
self.STEP_Z_SPAN,
self.leg_length * self.STEP_Z_MAX)
# Leg IK
a = self.act
l_ankle_pos = (a[0] * 0.025 - 0.01, a[1] * 0.01 + lfy, a[2] * 0.01 + lfz)
r_ankle_pos = (a[3] * 0.025 - 0.01, a[4] * 0.01 + rfy, a[5] * 0.01 + rfz)
l_foot_rot = a[6:9] * (2, 2, 3)
r_foot_rot = a[9:12] * (2, 2, 3)
# Limit leg yaw/pitch (and add bias)
l_foot_rot[2] = max(0, l_foot_rot[2] + 18.3)
r_foot_rot[2] = min(0, r_foot_rot[2] - 18.3)
# Arms actions
arms = np.copy(self.DEFAULT_ARMS) # default arms pose
arms[0:4] += a[12:16] * 4 # arms pitch+roll
# Set target positions
self.execute_ik(l_ankle_pos, l_foot_rot, r_ankle_pos, r_foot_rot) # legs
r.set_joints_target_position_direct(slice(14, 22), arms, harmonize=False) # arms
self.sync()
self.step_counter += 1
# 收集观测
obs = self.observe()
angle_rad = np.radians(self.gym_last_internal_abs_ori) # 将角度转换为弧度
unit_vector = np.array([np.cos(angle_rad), np.sin(angle_rad)]) # 计算单位向量
if np.linalg.norm(w.ball_cheat_abs_vel[:2]) != 0:
cos_theta = np.dot(unit_vector, w.ball_cheat_abs_vel[:2]) / (
np.linalg.norm(unit_vector) * np.linalg.norm(w.ball_cheat_abs_vel[:2]))
else:
cos_theta = 0
# 计算奖励
reward = np.linalg.norm(w.ball_cheat_abs_vel) * cos_theta
if self.ball_dist_hip_center_2d < 0.115:
reward = 0
# 判断终止
if self.player.behavior.is_ready("Get_Up"):
self.terminal = True
elif w.time_local_ms - self.reset_time > 7500 * 3 or np.linalg.norm(
w.ball_cheat_abs_pos[:2] - r.loc_head_position[:2]) > 0.45 or not w.is_ball_abs_pos_from_vision:
self.terminal = True
else:
self.terminal = False
return obs, reward, self.terminal, {}
class Train(Train_Base):
def __init__(self, script) -> None:
super().__init__(script)
def load_expert_data(self, path: str):
# 加载专家数据
expert_data = np.load(path, allow_pickle=True)
observations, actions = [], []
for item in expert_data:
if isinstance(item, dict):
# 处理字典格式的数据
observations.append(item['observations'])
actions.append(item['actions'])
elif isinstance(item, tuple) and len(item) == 2:
# 处理元组格式的数据
observations.append(item[0])
actions.append(item[1])
else:
print("Unexpected item format:", item)
# 将列表转换为numpy数组
return {
'observations': np.array(observations, dtype=np.float32),
'actions': np.array(actions, dtype=np.float32)
}
def train(self, args):
# --------------------------------------- 学习参数
n_envs = min(1, os.cpu_count())
n_steps_per_env = 1024 # RolloutBuffer is of size (n_steps_per_env * n_envs)
minibatch_size = 64 # should be a factor of (n_steps_per_env * n_envs)
learning_rate = 3e-4
total_steps = 50000000
folder_name = f'Sprint_Gail_R{self.robot_type}'
model_path = f'./scripts/gyms/logs/{folder_name}/'
if not os.path.exists(model_path):
os.makedirs(model_path)
# 加载专家数据
expert_data = self.load_expert_data(
"/home/her-darling/APOLLO_PRO/TEST_Dribble/scripts/gyms/dribble_expert_data.pkl")
# --------------------------------------- 运行 GAIL 算法
def init_env(i_env):
def thunk():
return dribble(self.ip, self.server_p + i_env, self.monitor_p_1000 + i_env, self.robot_type, False)
return thunk
servers = Server(self.server_p, self.monitor_p_1000, n_envs + 1) # include 1 extra server for testing
env = SubprocVecEnv([init_env(i) for i in range(n_envs)])
eval_env = SubprocVecEnv([init_env(n_envs)])
reward_net = BasicRewardNet(observation_space=env.observation_space, action_space=env.action_space)
try:
if "model_file" in args: # retrain
model = PPO.load(args["model_file"], env=env, device="cuda", 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="cuda")
gail_trainer = GAIL(
demonstrations=expert_data,
demo_batch_size=32,
venv=env,
gen_algo=model, # 这里传递 PPO 实例,而不是字符串
reward_net=reward_net,
)
eval_callback = EvalCallback(
eval_env,
best_model_save_path='./logs/',
log_path='./logs/',
eval_freq=10000, # 每训练 10000 个 step 进行一次评估
deterministic=True,
render=False,
)
gail_trainer.train(total_timesteps=total_steps, callback=eval_callback)
except KeyboardInterrupt:
sleep(1) # 等待子进程结束
print("\n按下 ctrl+c正在中止...\n")
servers.kill()
return
except Exception as e:
print(f"训练过程中出错: {e}")
env.close()
eval_env.close()
servers.kill()
def test(self, args):
server = Server(self.server_p - 1, self.monitor_p, 1)
env = dribble(self.ip, self.server_p - 1, self.monitor_p, self.robot_type, True)
model = GAIL.load(args["model_file"], env=env) # 加载 GAIL 模型
try:
self.export_model(args["model_file"], args["model_file"] + ".pkl", False)
self.test_model(model, env, log_path=args["folder_dir"], model_path=args["folder_dir"])
except KeyboardInterrupt:
print()
env.close()
server.kill()

@ -0,0 +1,403 @@
import math
import pickle
import random
from agent.Base_Agent import Base_Agent as Agent
from behaviors.custom.Step.Step import Step
from world.commons.Draw import Draw
from stable_baselines3 import PPO
from stable_baselines3.common.vec_env import SubprocVecEnv
from scripts.commons.Server import Server
from scripts.commons.Train_Base import Train_Base
from time import sleep
import os, gym
import numpy as np
from math_ops.Math_Ops import Math_Ops as M
from behaviors.custom.Step.Step_Generator import Step_Generator
'''
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 dribble(gym.Env):
def __init__(self, ip, server_p, monitor_p, r_type, enable_draw) -> None:
self.abs_ori = 75
self.ball_dist_hip_center_2d = 0
self.ball_dist_hip_center = None
self.internal_rel_orientation = None
self.dribble_speed = 1
self.gym_last_internal_abs_ori = None
self.internal_target_vel = None
self.Gen_player_pos = None
self.internal_target = None
self.values_l = None
self.values_r = None
self.reset_time = None
self.behavior = None
self.robot_type = r_type
self.kick_ori = 0
self.terminal = False
# Args: Server IP, Agent Port, Monitor Port, Uniform No., Robot Type, Team Name, Enable Log, Enable Draw
self.player = Agent(ip, server_p, monitor_p, 1, self.robot_type, "Gym", True, True)
self.step_counter = 0 # to limit episode size
self.ik = self.player.inv_kinematics
self.dribble_rel_orientation = 0 # relative to imu_torso_orientation (in degrees)
# Step behavior defaults
self.STEP_DUR = 8
self.STEP_Z_SPAN = 0.02
self.STEP_Z_MAX = 0.70
nao_specs = self.ik.NAO_SPECS
self.leg_length = nao_specs[1] + nao_specs[3] # upper leg height + lower leg height
feet_y_dev = nao_specs[0] * 1.12 # wider step
sample_time = self.player.world.robot.STEPTIME
max_ankle_z = nao_specs[5]
self.step_generator = Step_Generator(feet_y_dev, sample_time, max_ankle_z)
self.DEFAULT_ARMS = np.array([-90, -90, 8, 8, 90, 90, 70, 70], np.float32)
self.act = np.zeros(16, np.float32) # memory variable
self.path_manager = self.player.path_manager
# State space
obs_size = 76
self.obs = np.zeros(obs_size, np.float32)
self.observation_space = gym.spaces.Box(low=np.full(obs_size, -np.inf, np.float32),
high=np.full(obs_size, np.inf, np.float32), dtype=np.float32)
self.ball_x_center = 0.21
self.ball_y_center = -0.045
# Action space
MAX = np.finfo(np.float32).max
self.no_of_actions = act_size = 16
self.action_space = gym.spaces.Box(low=np.full(act_size, -MAX, np.float32),
high=np.full(act_size, MAX, np.float32), dtype=np.float32)
# Place ball far away to keep landmarks in FoV (head follows ball while using Step behavior)
self.player.scom.unofficial_move_ball((14, 0, 0.042))
self.player.scom.unofficial_set_play_mode("PlayOn")
self.player.scom.unofficial_move_ball((0, 0, 0))
with open("/home/her-darling/APOLLO_PRO/TEST_Dribble/scripts/gyms/dribble_observations.pkl", 'rb') as f:
self.obs_list = pickle.load(f)
self.obs_list = np.array(self.obs_list)
def observe(self, init=False, virtual_ball=False):
r = self.player.world.robot
w = self.player.world
if init: # reset variables
self.step_counter = 0
self.act = np.zeros(16, np.float32) # memory variable
# index observation naive normalization
self.obs[0] = min(self.step_counter, 12 * 8) / 100 # simple counter: 0,1,2,3...
self.obs[1] = r.loc_head_z * 3 # z coordinate (torso)
self.obs[2] = r.loc_head_z_vel / 2 # z velocity (torso)
self.obs[3] = r.imu_torso_roll / 15 # absolute torso roll in deg
self.obs[4] = r.imu_torso_pitch / 15 # absolute torso pitch in deg
self.obs[5:8] = r.gyro / 100 # gyroscope
self.obs[8:11] = r.acc / 10 # accelerometer
self.obs[11:17] = r.frp.get('lf', np.zeros(6)) * (10, 10, 10, 0.01, 0.01,
0.01) # left foot: relative point of origin (p) and force vector (f) -> (px,py,pz,fx,fy,fz)*
self.obs[17:23] = r.frp.get('rf', np.zeros(6)) * (10, 10, 10, 0.01, 0.01,
0.01) # right foot: relative point of origin (p) and force vector (f) -> (px,py,pz,fx,fy,fz)*
# *if foot is not touching the ground, then (px=0,py=0,pz=0,fx=0,fy=0,fz=0)
self.obs[23:43] = r.joints_position[2:22] / 100 # position of all joints except head & toes (for robot type 4)
self.obs[43:63] = r.joints_speed[2:22] / 6.1395 # speed of all joints except head & toes (for robot type 4)
if init: # the walking parameters refer to the last parameters in effect (after a reset, they are pointless)
self.obs[63] = 1 # step progress
self.obs[64] = 1 # 1 if left leg is active
self.obs[65] = 0 # 1 if right leg is active
self.obs[66] = 0
else:
self.obs[63] = self.step_generator.external_progress # step progress
self.obs[64] = float(self.step_generator.state_is_left_active) # 1 if left leg is active
self.obs[65] = float(not self.step_generator.state_is_left_active) # 1 if right leg is active
self.obs[66] = math.sin(self.step_generator.state_current_ts / self.step_generator.ts_per_step * math.pi)
# Ball
ball_rel_hip_center = self.ik.torso_to_hip_transform(w.ball_rel_torso_cart_pos)
self.ball_dist_hip_center = np.linalg.norm(ball_rel_hip_center)
ball_rel_torso_xy = w.ball_rel_torso_cart_pos[:2] # 取X和Y分量
self.ball_dist_hip_center_2d = np.linalg.norm(ball_rel_torso_xy)
if init:
self.obs[67:70] = (0, 0, 0) # Initial velocity is 0
elif w.ball_is_visible:
self.obs[67:70] = (ball_rel_hip_center - self.obs[
70:73]) * 10 # Ball velocity, relative to ankle's midpoint
self.obs[70:73] = ball_rel_hip_center # Ball position, relative to hip
self.obs[73] = self.ball_dist_hip_center * 2
if virtual_ball: # simulate the ball between the robot's feet
self.obs[67:74] = (0, 0, 0, 0.05, 0, -0.175, 0.36)
'''
Create internal target with a smoother variation
'''
MAX_ROTATION_DIFF = 20 # max difference (degrees) per visual step
MAX_ROTATION_DIST = 80
if init:
self.internal_rel_orientation = 0
self.internal_target_vel = 0
self.gym_last_internal_abs_ori = r.imu_torso_orientation # for training purposes (reward)
# ---------------------------------------------------------------- compute internal target
if w.vision_is_up_to_date:
previous_internal_rel_orientation = np.copy(self.internal_rel_orientation)
internal_ori_diff = np.clip(M.normalize_deg(self.dribble_rel_orientation - self.internal_rel_orientation),
-MAX_ROTATION_DIFF, MAX_ROTATION_DIFF)
self.internal_rel_orientation = np.clip(M.normalize_deg(self.internal_rel_orientation + internal_ori_diff),
-MAX_ROTATION_DIST, MAX_ROTATION_DIST)
# Observations
self.internal_target_vel = self.internal_rel_orientation - previous_internal_rel_orientation
self.gym_last_internal_abs_ori = self.internal_rel_orientation + r.imu_torso_orientation
# ----------------------------------------------------------------- observations
self.obs[74] = self.internal_rel_orientation / MAX_ROTATION_DIST
self.obs[75] = self.internal_target_vel / MAX_ROTATION_DIFF
return self.obs
def execute_ik(self, l_pos, l_rot, r_pos, r_rot):
r = self.player.world.robot
# Apply IK to each leg + Set joint targets
# Left leg
indices, self.values_l, error_codes = self.ik.leg(l_pos, l_rot, True, dynamic_pose=False)
r.set_joints_target_position_direct(indices, self.values_l, harmonize=False)
# Right leg
indices, self.values_r, error_codes = self.ik.leg(r_pos, r_rot, False, dynamic_pose=False)
r.set_joints_target_position_direct(indices, self.values_r, harmonize=False)
def sync(self):
''' Run a single simulation step '''
r = self.player.world.robot
self.player.scom.commit_and_send(r.get_command())
self.player.scom.receive()
def reset(self):
# print("reset")
'''
Reset and stabilize the robot
Note: for some behaviors it would be better to reduce stabilization or add noise
'''
self.abs_ori = 45
self.player.scom.unofficial_set_play_mode("PlayOn")
Gen_ball_pos = [- 9, 0, 0]
self.Gen_player_pos = (Gen_ball_pos[0] - 1.5, Gen_ball_pos[1], 0.5)
self.player.scom.unofficial_move_ball((Gen_ball_pos[0], Gen_ball_pos[1], Gen_ball_pos[2]))
self.step_counter = 0
self.behavior = self.player.behavior
r = self.player.world.robot
w = self.player.world
t = w.time_local_ms
self.reset_time = t
self.b_rel = w.ball_rel_torso_cart_pos[:2]
self.path_manager = self.player.path_manager
for _ in range(25):
self.player.scom.unofficial_beam(self.Gen_player_pos, 0) # beam player continuously (floating above ground)
self.player.behavior.execute("Zero_Bent_Knees")
self.sync()
# beam player to ground
self.player.scom.unofficial_beam(self.Gen_player_pos, 0)
r.joints_target_speed[
0] = 0.01 # move head to trigger physics update (rcssserver3d bug when no joint is moving)
self.sync()
# stabilize on ground
for _ in range(7):
self.player.behavior.execute("Zero_Bent_Knees")
self.sync()
# walk to ball
while True:
self.b_rel = w.ball_rel_torso_cart_pos[:2]
if self.player.behavior.is_ready("Get_Up"):
self.player.behavior.execute_to_completion("Get_Up")
if 0.26 > self.b_rel[0] > 0.18 and abs(self.b_rel[1]) < 0.04 and w.ball_is_visible:
break
else:
if self.player.behavior.is_ready("Get_Up"):
self.player.behavior.execute_to_completion("Get_Up")
reset_walk = self.behavior.previous_behavior != "Walk" # reset walk if it wasn't the previous behavior
rel_target = self.b_rel + (-0.23, 0) # relative target is a circle (center: ball, radius:0.23m)
rel_ori = M.vector_angle(self.b_rel) # relative orientation of ball, NOT the target!
dist = max(0.08, np.linalg.norm(rel_target) * 0.7) # slow approach
self.behavior.execute_sub_behavior("Walk", reset_walk, rel_target, False, rel_ori, False,
dist) # target, is_target_abs, ori, is_ori_abs, distance
self.sync()
self.act = np.zeros(self.no_of_actions, np.float32)
return self.observe(True)
def render(self, mode='human', close=False):
return
def close(self):
Draw.clear_all()
self.player.terminate()
def step(self, action):
r = (self.
player.world.robot)
w = self.player.world
d = w.draw
if w.ball_abs_pos[1] > 0: #
dribble_target = (15, 5)
else:
dribble_target = (15, -5)
self.dribble_rel_orientation = self.path_manager.get_dribble_path(optional_2d_target=dribble_target)[1]
# exponential moving average
self.act = 0.85 * self.act + 0.15 * action * 0.7 * 0.95 * self.dribble_speed
# execute Step behavior to extract the target positions of each leg (we will override these targets)
lfy, lfz, rfy, rfz = self.step_generator.get_target_positions(self.step_counter == 0, self.STEP_DUR,
self.STEP_Z_SPAN,
self.leg_length * self.STEP_Z_MAX)
# Leg IK
a = self.act
l_ankle_pos = (a[0] * 0.025 - 0.01, a[1] * 0.01 + lfy, a[2] * 0.01 + lfz)
r_ankle_pos = (a[3] * 0.025 - 0.01, a[4] * 0.01 + rfy, a[5] * 0.01 + rfz)
l_foot_rot = a[6:9] * (2, 2, 3)
r_foot_rot = a[9:12] * (2, 2, 3)
# Limit leg yaw/pitch (and add bias)
l_foot_rot[2] = max(0, l_foot_rot[2] + 18.3)
r_foot_rot[2] = min(0, r_foot_rot[2] - 18.3)
# Arms actions
arms = np.copy(self.DEFAULT_ARMS) # default arms pose
arms[0:4] += a[12:16] * 4 # arms pitch+roll
# Set target positions
self.execute_ik(l_ankle_pos, l_foot_rot, r_ankle_pos, r_foot_rot) # legs
r.set_joints_target_position_direct(slice(14, 22), arms, harmonize=False) # arms
self.sync()
self.step_counter += 1
# 收集观测
obs = self.observe()
angle_rad = np.radians(self.gym_last_internal_abs_ori) # 将角度转换为弧度
unit_vector = np.array([np.cos(angle_rad), np.sin(angle_rad)]) # 计算单位向量
if np.linalg.norm(w.ball_cheat_abs_vel[:2]) != 0:
cos_theta = np.dot(unit_vector, w.ball_cheat_abs_vel[:2]) / (
np.linalg.norm(unit_vector) * np.linalg.norm(w.ball_cheat_abs_vel[:2]))
else:
cos_theta = 0
# 计算奖励
ref = self.obs_list[self.step_counter + 3]
loss = self.cal_reward(ref, self.obs) * 0.075
reward = np.linalg.norm(w.ball_cheat_abs_vel) * cos_theta + loss
if self.ball_dist_hip_center_2d < 0.115:
reward = 0
# 判断终止
if self.player.behavior.is_ready("Get_Up"):
self.terminal = True
elif w.time_local_ms - self.reset_time > 7500 * 3 or np.linalg.norm(
w.ball_cheat_abs_pos[:2] - r.loc_head_position[:2]) > 0.45 or not w.is_ball_abs_pos_from_vision:
self.terminal = True
else:
self.terminal = False
return obs, reward, self.terminal, {}
def cal_reward(self,ref,obs):
reward = -0.5*np.linalg.norm(ref[23:39]*100 - obs[23:39]*100) -0.5*np.linalg.norm(ref[1:23]*30 - obs[1:23]*30)
reward -= (2*np.linalg.norm(ref[20:23] - obs[20:23]) + 2*np.linalg.norm(ref[14:17] - obs[14:17]))
return reward
class Train(Train_Base):
def __init__(self, script) -> None:
super().__init__(script)
def train(self, args):
# --------------------------------------- Learning parameters
n_envs = min(12, os.cpu_count())
n_steps_per_env = 1024 # RolloutBuffer is of size (n_steps_per_env * n_envs)
minibatch_size = 64 # should be a factor of (n_steps_per_env * n_envs)
total_steps = 50000000
learning_rate = 3e-4
folder_name = f'Sprint_mimic_R{self.robot_type}'
model_path = f'./scripts/gyms/logs/{folder_name}/'
# print("Model path:", model_path)
# --------------------------------------- Run algorithm
def init_env(i_env):
def thunk():
return dribble(self.ip, self.server_p + i_env, self.monitor_p_1000 + i_env, self.robot_type, False)
return thunk
servers = Server(self.server_p, self.monitor_p_1000, n_envs + 1) # include 1 extra server for testing
env = SubprocVecEnv([init_env(i) for i in range(n_envs)])
eval_env = SubprocVecEnv([init_env(n_envs)])
try:
if "model_file" in args: # retrain
model = PPO.load(args["model_file"], env=env, device="cuda", 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="cuda")
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
("\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 = Server(self.server_p - 1, self.monitor_p, 1)
env = dribble(self.ip, self.server_p - 1, self.monitor_p, self.robot_type, True)
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()
Loading…
Cancel
Save