-
Notifications
You must be signed in to change notification settings - Fork 0
/
environment.py
300 lines (249 loc) · 9.82 KB
/
environment.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
import copy
import mujoco
import mujoco.viewer
import numpy as np
from Bezier import get_my_Bezier_point,pos_2_angle
import math
import random
import torch
import matplotlib.pyplot as plt
delta_St = np.array([0, 1, 1, 0])
# delta_St = np.array([0, 0.5, 0.5, 0])
# delta_St = np.array([0, 0.2, 0.55, 0.75])
# delta_St = np.array([0, 0.4, 1.1, 1.5])
# delta_St = np.array([0, 0.5, 1.0, 1.5])
dt = 0.005
# dt = 0.01
# dt = 0.015
# dt = 0.02
# dt = 0.025
# dt = 0.07
# dt = 0.09
# dt = 0.011
# dt = 0.013
para = [[[-0.00, -0.045], [0.03, 0.01]], [[-0.00, -0.045], [0.03, 0.005]], [[0.00, -0.05], [0.03, 0.01]],[[0.00, -0.05], [0.03, 0.005]]]
max_dis=0.03
class Env(object):
def __init__(self, modelPath, max_steps=5000, view=False):
super().__init__()
self.model = mujoco.MjModel.from_xml_path(modelPath) # 加载模型
self.data = mujoco.MjData(self.model)
self.render = view
if self.render:
self.viewer = mujoco.viewer.launch_passive(self.model, self.data)
else:
self.viewer = None
self.legPosName = [
# shoulder对应front left/right leg(fl/fr)，hip对应rear left/right leg(rl/rr)；router为上方关节、foot为下方关节
["router_shoulder_fl", "foot_s_fl"],
["router_shoulder_fr", "foot_s_fr"],
["router_hip_rl", "foot_s_rl"],
["router_hip_rr", "foot_s_rr"]]
self.fixPoint = "body_ss" # "neck_ss"
self.index = {"m1_fl": 0, "m2_fl": 1, "m1_fr": 2, "m2_fr": 3, "m1_rl": 4, "m2_rl": 5, "m1_rr": 6, "m2_rr": 7,
"m1_tail": 8,
"neck": 9, "head": 10, "spine": 11, "fl_t1": 12, "fr_t1": 13, "rl_t1": 14, "rr_t1": 15,
"com_pos": 16, "com_quat": 19, "com_vel": 23, "imu_acc": 26, "imu_gyro": 29} # 设置传感器的序号
self.movePath = [[], [], []]
self.control_point_list = []
self.pos = []
self.linvel = []
self.angvel = []
self.acc = []
self.last_pos = [0, 0, 0]
self.last_action = 0
self.max_steps = max_steps
self.steps = 0
self.ctrldata = np.array([])
self.t=0
self.old_ctrl=0
self.n=0
self.initializing()
def initializing(self):
#初始化
for i in range(50):
self.data.ctrl = [0.0, 1.5, 0.0, 1.5, 0.0, -1.2, 0.0, -1.2, 0, 0, 0, 0]
mujoco.mj_step(self.model, self.data)
if self.render:
self.viewer.sync()
# self.random_env()
self.old_ctrl=0
self.t=0
self.steps = 0
self.ctrldata = np.array([])
state, pos, vel,acc = self.get_sensors()
self.last_pos = copy.deepcopy(pos)
self.last_action = 0
self.n=0
self.movePath = [[], [], []]
self.control_point_list = []
self.pos = []
self.linvel = []
self.angvel = []
self.acc = []
# quat=state[-4:]
# rm=quat2rm(quat)
state = np.concatenate([[-1]*12, state, self.t + delta_St, [0]])
# state = np.concatenate([[-1]*10, state, self.t + delta_St, [0]])
return state
def runStep(self, ctrlData):
# ------------------------------------------ #
# ID 0, 1 left-fore leg and coil
# ID 2, 3 right-fore leg and coil
# ID 4, 5 left-hide leg and coil
# ID 6, 7 right-hide leg and coil
# Note: For leg, it has [-1: front; 1: back]
# Note: For fore coil, it has [-1: leg up; 1: leg down]
# Note: For hide coil, it has [-1: leg down; 1: leg up]
# ------------------------------------------ #
# ID 08 is neck (Horizontal)
# ID 09 is head (vertical)
# ID 10 is spine (Horizontal) [-1: right, 1: left]
# Note: range is [-1, 1]
# ------------------------------------------ #
ctrlData*=max_dis
ctrlData[0]=ctrlData[0]-max_dis
ctrlData[1]=(ctrlData[1]-max_dis)*1.4
ctrlData[2]=(ctrlData[2]-max_dis)*1.5
ctrlData[3]*=0.5
ctrlData[4]=(ctrlData[4]+max_dis)*1.5
ctrlData[5]=(ctrlData[5]+max_dis)*1.4
ctrlData[6]=ctrlData[6]+max_dis
ctrlData[7]=(ctrlData[7]+max_dis)*0.3
ctrlData[8]=(ctrlData[8]+max_dis)*0.3
ctrlData[9]=(ctrlData[9]+max_dis)*0.4
ctrlData[10]=ctrlData[10]+max_dis
ctrlData[11]=(ctrlData[11]-max_dis)*0.2
# ctrlData=0.3*self.old_ctrl+0.7*ctrlData
self.old_ctrl=copy.deepcopy(ctrlData)
old=np.array(self.data.ctrl[:8])
for i in range(4):
x, y, control_points = get_my_Bezier_point(ctrlData, self.t + delta_St[i])
if i < 2:
leg = "f"
else:
leg = "h"
dx = para[i][0][0]
dy = para[i][0][1]
X = x + dx
Y = y + dy
self.data.ctrl[i * 2:(i + 1) * 2] = pos_2_angle(X, Y, leg)
self.data.ctrl[8:]=0
now=np.array(self.data.ctrl[:8])
dif=np.mean(np.abs(now-old))
mujoco.mj_step(self.model, self.data)
self.steps+=1
if self.render:
self.viewer.sync()
tData = self.data.site(self.fixPoint).xpos
state,pos,linvel,acc = self.get_sensors()
for i in range(3):
self.movePath[i].append(tData[i])
self.control_point_list.append(control_points)
self.pos.append(pos)
self.linvel.append(linvel)
self.angvel.append(state[-7:-4])
self.acc.append(acc)
c_F=np.sum(state[:4])
if c_F ==0:
self.n+=1
else:
self.n=0
# quat=state[-4:]
# rm=quat2rm(quat)
if abs(linvel[1])<0.35:
reward = -linvel[1] * 4
else:
reward = - 0.1
# reward-=0.1*dif
# if self.n>10:
# reward-=0.1
self.t = (self.t + dt) % 2
state = np.concatenate([ctrlData, state, self.t + delta_St, [reward]])
done=False
if self.steps >= self.max_steps:
done = True
self.last_pos=copy.deepcopy(pos)
return state, reward, done, pos
def get_sensors(self): # 得到观测值与质心位置坐标
sensors = self.data.sensordata
pos = sensors[self.index["com_pos"]:self.index["com_pos"] + 3].copy() # 位置
acc = sensors[self.index["imu_acc"]:self.index["imu_acc"] + 3].copy() # 加速度
angvel = sensors[self.index["imu_gyro"]:self.index["imu_gyro"] + 3].copy() # 角速度
framequat = sensors[self.index["com_quat"]:self.index["com_quat"] + 4].copy() # 四元数（代替Pitch Angle（俯仰角）和Body Roll（滚转角））
linvel = sensors[self.index["com_vel"]:self.index["com_vel"] + 3].copy() # 线速度
c_F = sensors[self.index["fl_t1"]:self.index["fl_t1"] + 4].copy() # 腿部压力传感器
state = np.concatenate([c_F, linvel, angvel, framequat])
return state,pos,linvel,acc
def reset(self):
mujoco.mj_resetData(self.model, self.data)
return self.initializing()
def random_env(self):
# self.model.opt (可以修改如重力、风速之类的参数)
# self.model.opt.gravity[-1]=-5
# print(self.model.opt.gravity)
# self.model.opt.gravity[-1] = -12 + random.random() * 5
for i in range(1, 451):
self.model.geom("box{}".format(i)).size[-1] = random.random() * 0.01
def drawPath(self):
ax = plt.axes(projection='3d')
ax.plot3D(self.movePath[0], self.movePath[1], self.movePath[2])
ax.set_xlabel('X', fontsize=16)
ax.set_ylabel('Y', fontsize=16)
ax.set_zlabel('Z', fontsize=16)
plt.show()
def write_log(self): # 记录传感器数据
def compute_L2(arr,digit=4):
sum = 0
for i in range(len(arr)):
sum += arr[i] ** 2
arr[i] = round(arr[i], digit)
sum = round(math.sqrt(sum), digit)
return sum
f1 = open("position.txt", mode='w')
f2 = open("linear velocity.txt", mode='w')
f3 = open("angular velocity.txt", mode='w')
f4 = open("acceleration.txt", mode='w')
f5 = open("control point.txt",mode='w')
for i in range(len(self.pos)):
pos_scalar = compute_L2(self.pos[i])
linvel_scalar = compute_L2(self.linvel[i])
angvel_scalar = compute_L2(self.angvel[i])
acc_scalar = compute_L2(self.acc[i])
f1.writelines(str(self.pos[i]) + ": " + str(pos_scalar) + "\n")
f2.writelines(str(self.linvel[i]) + ": " + str(linvel_scalar) + "\n")
f3.writelines(str(self.angvel[i]) + ": " + str(angvel_scalar) + "\n")
f4.writelines(str(self.acc[i]) + ": " + str(acc_scalar) + "\n")
f5.writelines(str(self.control_point_list[i]) + "\n")
f1.close()
f2.close()
f3.close()
f4.close()
f5.close()
def generate_boxes(x_min=-0.5, x_max=0.5, y_min=-3.5, y_max=1, size=0.05):
with open("box.txt", 'w') as file:
x = x_min
counter = 1
while x_max - x > 1e-6:
y = y_min
while y_max - y > 1e-6:
file.write(
'<geom name="box{}" type="box" size="{:.3} {:.3} {:.3}" pos="{:.3} {:.3} 0" rgba="0.5 0.5 0.5 1"/> \n'.format(
counter, size, size, size, x, y))
y += size * 2
counter += 1
x += size * 2
def quat2rm(q):
r=np.zeros([3,3])
r[0,0]=q[0]**2+q[1]**2-q[2]**2-q[3]**2
r[0,1]=2*(q[1]*q[2]-q[0]*q[3])
r[0,2]=2*(q[1]*q[3]+q[0]*q[2])
r[1,0]=2*(q[1]*q[2]+q[0]*q[3])
r[1,1]=q[0]**2-q[1]**2+q[2]**2-q[3]**2
r[1,2]=2*(q[2]*q[3]-q[0]*q[1])
r[2,0]=2*(q[1]*q[3]-q[0]*q[2])
r[2,1]=2*(q[2]*q[3]+q[0]*q[1])
r[2,2]=q[0]**2-q[1]**2-q[2]**2+q[3]**2
return r
if __name__ == '__main__':
generate_boxes()