-
Notifications
You must be signed in to change notification settings - Fork 0
/
fig_plot_rl3.py
262 lines (234 loc) · 8.55 KB
/
fig_plot_rl3.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
#!/usr/bin/env python
# -*- encoding: utf-8 -*-
"""用于评估任务二的最优表现
Write detailed description here
Write typical usage example here
@Modify Time @Author @Version @Description
------------ ------- -------- -----------
5/15/23 12:22 PM yinzikang 1.0 None
"""
import gym
import matplotlib.pyplot as plt
import numpy as np
from gym_custom.envs.env_kwargs import env_kwargs
from gym_custom.utils.custom_loader import load_episode
from gym_custom.envs.controller import orientation_error_quat_with_quat, mat33_to_quat
from gym_custom.envs.transformations import quaternion_multiply, quaternion_inverse
plt.rcParams['font.family'] = 'Times New Roman'
plt.rcParams['font.size'] = 10.5
plt.rcParams['lines.linewidth'] = 2.0
env_name = 'TrainEnvVariableStiffnessAndPostureAndSM_v2-v8'
test_name = 'cabinet drawer open with plan'
rl_name = 'PPO'
# time_name = '05-07-22-47'
# # time_name = '05-08-09-09'
# path_name = test_name + '/' + rl_name + '/' + time_name + '/'
# logger_path = "eval_results/" + path_name + "best_model"
# save_dir = './figs/' + test_name
# 优化前
path_name = './gym_custom/rl_test_results/' + test_name + '/' + env_name + '/'
logger_path = path_name
save_dir = logger_path
_, _, rl_kwargs = env_kwargs(test_name, save_flag=False)
env = gym.make(env_name, **rl_kwargs)
result_dict = load_episode(logger_path)
save_flag = True
view_flag = True
duration = 2000
angle = np.pi / 60
r = np.sqrt(0.2175 ** 2 + 0.475 ** 2)
theta = np.arctan(0.475 / 0.2175) - angle
start_point = np.array([0.9 - r * np.cos(theta), -0.1135, 0.2 + r * np.sin(theta)]) + np.array([-0.011, 0, 0.004])
end_point = start_point + np.array([-0.3 * np.cos(angle), 0, - 0.3 * np.sin(angle)])
pos_table = np.linspace(start_point, end_point, num=duration - 1, axis=0)
table = np.array([[0.9986295, 0, -0.0523360],
[0, 1, 0],
[0.0523360, 0, 0.9986295]])
quat_table = mat33_to_quat(table)
i = 0
# 位置误差：实际运行result_dict["xpos"]、修正前env.init_desired_xposture_list[:, :3]、修正后result_dict["desired_xpos"]
# 对比修正后的期望轨迹与门的一致性
pos_init = env.init_desired_xposture_list[:duration - 1, 0:3]
pos_adjusted = result_dict["desired_xpos"]
pos_real = result_dict["xpos"]
plt.figure(i)
plt.plot(pos_table[:, 0], pos_table[:, 2], label='cabinet drawer')
plt.plot(pos_init[:, 0], pos_init[:, 2], label='desired path')
plt.plot(pos_adjusted[:, 0], pos_adjusted[:, 2], label='path adjusted by action')
plt.plot(pos_real[:, 0], pos_real[:, 2], label='actual path')
plt.legend(loc='lower right')
plt.xlabel(r'X position $\mathrm{(m)}$')
plt.ylabel(r'Z position $\mathrm{(m)}$')
plt.grid()
if save_flag:
plt.savefig(save_dir + '/drawer xpos.png', dpi=600, bbox_inches='tight')
if view_flag:
plt.title('drawer xpos')
xpos_error_table_buffer1 = []
xpos_error_table_buffer2 = []
for j in range(len(result_dict["xquat"])):
xpos_error_table_buffer1.append(table.transpose() @ (pos_real - pos_init)[j])
xpos_error_table_buffer2.append(table.transpose() @ (pos_real - pos_adjusted)[j])
xpos_error_table_buffer1 = np.array(xpos_error_table_buffer1)
xpos_error_table_buffer2 = np.array(xpos_error_table_buffer2)
i += 1
plt.figure(i)
plt.plot(xpos_error_table_buffer1[:, 0], label='x')
plt.plot(xpos_error_table_buffer1[:, 1], label='y')
plt.plot(xpos_error_table_buffer1[:, 2], label='z')
plt.legend(loc='upper right')
plt.xlabel('steps')
plt.ylabel(r'position error $\mathrm{(m)}$')
plt.xlim([0, 2000])
plt.grid()
if save_flag:
plt.savefig(save_dir + '/drawer xpos error.png', dpi=600, bbox_inches='tight')
if view_flag:
plt.title('drawer xpos error with init')
i += 1
plt.figure(i)
plt.plot(xpos_error_table_buffer2[:, 0], label='x')
plt.plot(xpos_error_table_buffer2[:, 1], label='y')
plt.plot(xpos_error_table_buffer2[:, 2], label='z')
plt.legend(loc='upper right')
plt.xlabel('steps')
plt.ylabel(r'position error $\mathrm{(m)}$')
plt.xlim([0, 2000])
plt.grid()
if save_flag:
plt.savefig(save_dir + '/drawer xpos error.png', dpi=600, bbox_inches='tight')
if view_flag:
plt.title('drawer xpos error with adjusted')
# 姿态误差
mat_init = env.init_desired_xposture_list[:duration - 1, 3:12]
quat_init = env.init_desired_xposture_list[:duration - 1, 12:16]
quat_adjusted = result_dict["desired_xquat"]
quat_real = result_dict["xquat"]
quat_optimal = quaternion_multiply(quat_table, quat_init[0])
orientation_error_buffer1 = []
orientation_error_buffer2 = []
orientation_error_buffer3 = []
for j in range(len(result_dict["xquat"])):
orientation_error_buffer1.append(
orientation_error_quat_with_quat(quat_init[j], quat_real[j]))
orientation_error_buffer2.append(
orientation_error_quat_with_quat(quat_adjusted[j], quat_real[j]))
orientation_error_buffer3.append(
orientation_error_quat_with_quat(quat_optimal, quat_real[j]))
orientation_error_buffer1 = np.array(orientation_error_buffer1)
orientation_error_buffer2 = np.array(orientation_error_buffer2)
orientation_error_buffer3 = np.array(orientation_error_buffer3)
i += 1
plt.figure(i)
plt.plot(orientation_error_buffer1[:, 0], label='x')
plt.plot(orientation_error_buffer1[:, 1], label='y')
plt.plot(orientation_error_buffer1[:, 2], label='z')
plt.legend(loc='upper right')
plt.xlabel('steps')
plt.ylabel(r'orientation error')
plt.xlim([0, 2000])
plt.grid()
if save_flag:
plt.savefig(save_dir + '/drawer xquat error.png', dpi=600, bbox_inches='tight')
if view_flag:
plt.title('orientation error with init')
i += 1
plt.figure(i)
plt.plot(orientation_error_buffer2[:, 0], label='x')
plt.plot(orientation_error_buffer2[:, 1], label='y')
plt.plot(orientation_error_buffer2[:, 2], label='z')
plt.legend(loc='upper right')
plt.xlabel('steps')
plt.ylabel(r'orientation error')
plt.xlim([0, 2000])
plt.grid()
if save_flag:
plt.savefig(save_dir + '/drawer xquat error.png', dpi=600, bbox_inches='tight')
if view_flag:
plt.title('orientation error with adjusted')
i += 1
plt.figure(i)
plt.plot(orientation_error_buffer3[:, 0], label='x')
plt.plot(orientation_error_buffer3[:, 1], label='y')
plt.plot(orientation_error_buffer3[:, 2], label='z')
plt.legend(loc='upper right')
plt.xlabel('steps')
plt.ylabel(r'orientation error')
plt.xlim([0, 2000])
plt.grid()
if save_flag:
plt.savefig(save_dir + '/drawer xquat error.png', dpi=600, bbox_inches='tight')
if view_flag:
plt.title('orientation error with optimal')
# 接触力
contact_force_table = np.empty_like(result_dict["contact_force"])
for k in range(contact_force_table.shape[0]):
contact_force_table[k] = (table.transpose() @ result_dict["contact_force"][k].reshape((3, 2), order="F")). \
reshape(-1, order="F")
i += 1
plt.figure(i)
plt.plot(contact_force_table[:, 0], label='x')
plt.plot(contact_force_table[:, 1], label='y')
plt.plot(contact_force_table[:, 2], label='z')
plt.legend(loc='upper right')
plt.xlabel('steps')
plt.ylabel(r'contact force $\mathrm{(N)}$')
plt.xlim([0, 2000])
plt.grid()
if save_flag:
plt.savefig(save_dir + '/drawer contact force table.png', dpi=600, bbox_inches='tight')
if view_flag:
plt.title('drawer contact force table')
# 刚度大小
i += 1
plt.figure(i)
plt.plot(result_dict["K"][:, 0], label='x')
plt.plot(result_dict["K"][:, 1], label='y')
plt.plot(result_dict["K"][:, 2], label='z')
plt.legend(loc='upper right')
plt.xlabel('steps')
plt.ylabel(r'stiffness $\mathrm{(N/m)}$')
plt.xlim([0, 2000])
plt.grid()
if save_flag:
plt.savefig(save_dir + '/drawer K.png', dpi=600, bbox_inches='tight')
if view_flag:
plt.title('drawer K')
# 刚度姿态
K_quat = []
for j in range(len(result_dict["xquat"])):
K_quat.append(mat33_to_quat(result_dict["direction"][k].reshape((3, 3))))
K_quat = np.array(K_quat)
i += 1
plt.figure(i)
plt.plot([0, 1999], [quat_table[0], quat_table[0]], label='dx')
plt.plot([0, 1999], [quat_table[1], quat_table[1]], label='dy')
plt.plot([0, 1999], [quat_table[2], quat_table[2]], label='dz')
plt.plot([0, 1999], [quat_table[3], quat_table[3]], label='dw')
plt.plot(K_quat[:, 0], label='x')
plt.plot(K_quat[:, 1], label='y')
plt.plot(K_quat[:, 2], label='z')
plt.plot(K_quat[:, 3], label='w')
plt.legend(loc='upper right')
plt.xlabel('steps')
plt.ylabel(r'stiffness quat')
plt.xlim([0, 2000])
plt.grid()
if save_flag:
plt.savefig(save_dir + '/drawer K quat.png', dpi=600, bbox_inches='tight')
if view_flag:
plt.title('drawer K quat')
# 奖励
i += 1
plt.figure(i)
plt.plot(result_dict["reward"])
plt.xlabel('steps')
plt.ylabel('average return')
plt.xlim([0, 80])
plt.grid()
if save_flag:
plt.savefig(save_dir + '/drawer reward.png', dpi=600, bbox_inches='tight')
if view_flag:
plt.title('drawer reward')
if view_flag:
plt.show()