-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathsample.py
244 lines (201 loc) · 8.86 KB
/
sample.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
import numpy as np
from copy import deepcopy
import time
from igibson.envs.igibson_env import iGibsonEnv
from matplotlib import pyplot as plt
import os
import yaml
import json
dt = [
np.array([
[np.cos(-0.08334 * np.pi), -np.sin(-0.08334 * np.pi), 0, 0],
[np.sin(-0.08334 * np.pi), np.cos(-0.08334 * np.pi), 0, 0],
[0, 0, 1, 0.007],
[0, 0, 0, 1]
]),
np.array([
[np.cos(0.08334 * np.pi), -np.sin(0.08334 * np.pi), 0, 0],
[np.sin(0.08334 * np.pi), np.cos(0.08334 * np.pi), 0, 0],
[0, 0, 1, 0.007],
[0, 0, 0, 1]
]),
]
def Quat2Rotation(x,y,z,w):
l1 = np.stack([1 - 2 * y ** 2 - 2 * z ** 2, 2 * x * y + 2 * w * z, 2 * x * z - 2 * w * y],axis=0)
l2 = np.stack([2 * x * y - 2 * w * z, 1 - 2 * x ** 2 - 2 * z ** 2, 2 * y * z + 2 * w * x],axis=0)
l3 = np.stack([2 * x * z + 2 * w * y, 2 * y * z - 2 * w * x, 1 - 2 * x ** 2 - 2 * y ** 2], axis=0)
T_w = np.stack([l1,l2,l3],axis=0)
return T_w
def Rotation2Quat(pose):
m11,m22,m33 = pose[0][0],pose[1][1],pose[2][2]
m12,m13,m21,m23,m31,m32 = pose[0][1],pose[0][2],pose[1][0],pose[1][2],pose[2][0],pose[2][1]
x,y,z,w = np.sqrt(m11-m22-m33+1)/2,np.sqrt(-m11+m22-m33+1)/2,np.sqrt(-m11-m22+m33+1)/2,np.sqrt(m11+m22+m33+1)/2
Quat_ = np.array([
[x,(m12+m21)/(4*x),(m13+m31)/(4*x),(m23-m32)/(4*x)],
[(m12+m21)/(4*y),y,(m23+m32)/(4*y),(m31-m13)/(4*y)],
[(m13 + m31) / (4 * z), (m23 + m32) / (4 * z), z,(m12 - m21) / (4 * z)],
[(m23 - m32) / (4 * w), (m31 - m13) / (4 * w), (m12 - m21) / (4 * w),w]
], dtype=np.float32)
index = np.array([x,y,z,w]).argmax()
Quat = Quat_[index]
return Quat
def control(theta, scan):
if 0.458334 <= theta <= 0.541667: # 0.41667 0.58334
act = 0
elif -0.5 <= theta < 0.458334:
act = 2
else:
act = 1
if not hasattr(control, 'turn'):
control.turn = -1
control.turn_count = 0
control.stright_count = 0
# 如果距离障碍物过近或control.turn > 0
if scan.min()*5.6 < 0.3 or control.turn > 0:
# print('control')
if control.turn < 0:
index = scan.argmin()
index_ = scan.argmax()
portion = (index/len(scan)) * 90
portion = portion//15
por_ = (index_/len(scan)) * 90
por_ = por_/15
control.turn = 2 if por_ < 3 else 1
control.turn_count = portion + 2
control.stright_count = 2
act = control.turn
else:
if not control.turn_count <= 0:
control.turn_count -= 1
act = control.turn
elif not control.stright_count <= 0:
control.stright_count -= 1
act = 0
else:
control.turn = -1
return act
def step(env, act, robot_T):
if act == 0:
state, reward, done, info = env.step(np.array([0.7, 0]))
elif act == 1 or act == 2:
T = np.matmul(robot_T, dt[act - 1])
pos = T[:3, -1]
ori = Rotation2Quat(T[:3, :3])
env.robots[0].set_position_orientation(pos, ori)
state, reward, done, info = env.step(np.array([0.0, 0.0]))
return state, reward, done, info
# 保存数据
def save(count, sc, observation, target_theta, action, robot_pos_ori, dist, pathdist):
if not os.path.exists('dataset'):
os.mkdir('dataset')
if not os.path.exists('dataset/'+sc):
os.mkdir('dataset/'+sc)
if not os.path.exists('dataset/'+sc+'/'+'%04i'%count):
os.mkdir('dataset/'+sc+'/'+'%04i'%count)
root = 'dataset/'+sc+'/'+'%04i'%count # 将整型count转为四位字符串
observation.tofile(os.path.join(root, 'observation.npy'))
target_theta.tofile(os.path.join(root, 'target_theta.npy'))
action.tofile(os.path.join(root, 'action.npy'))
robot_pos_ori.tofile(os.path.join(root, 'robot_pos_ori.npy'))
with open(os.path.join(root, 'info.txt'), 'w') as f:
f.write('distance:%f pathdist:%f'%(dist, pathdist))
# 数据集的目录
root = './navigation_scenarios/waypoints/full+'
with open('scenes_for_training.txt', 'r') as f:
scene_list = f.readlines()
scene = []
for sc in scene_list:
scene.append(sc[:-1])
# 遍历场景
for sc in scene:
with open(os.path.join(root, sc+'.json'), 'r') as f:
content = json.load(f)
config_data = yaml.load(open("turtlebot_nav.yaml", "r"), Loader=yaml.FullLoader)
env = iGibsonEnv(config_file=deepcopy(config_data), scene_id=sc, mode='headless', action_timestep=2 / 5)
count = 0
for episode in content:
observation = []
action = []
target_theta = []
robot_pos_ori = []
flag = False
state = env.reset()
# 起始位置
start_pos = np.array([episode['startX'], episode['startY'], episode['startZ']])
# 起始位姿
start_ori = np.array([0, 0, -np.sin(episode['startAngle']/2), np.cos(episode['startAngle']/2)])
# 终止位置
end_pos = np.array([episode['goalX'], episode['goalY'], episode['goalZ']])
# robot初始化
env.robots[0].set_position_orientation(start_pos, start_ori)
state,_,_,_ = env.step(np.array([0,0]))
episode['waypoints'].append([episode['goalX'], episode['goalY'], episode['goalZ']])
waypoints = np.array(episode['waypoints'])
collison_num = 0
map = deepcopy(env.scene.floor_map[episode['level']])/255
step_num=0
# 遍历路点中的位置
for pos in waypoints:
target_pos = np.array([pos[0], pos[1], pos[2], 1])
collison_count = 0
while True:
robot_pos = env.robots[0].get_position()
xxx,yyy = env.scene.world_to_map(robot_pos[:2])
map[xxx,yyy] = 3
plt.imshow(map)
plt.pause(0.01)
plt.clf()
# 若与当前的目标路点距离较近，则结束对当前目标路点的导航
if np.linalg.norm(robot_pos[:2] - target_pos[:2], ord=2) <= 0.2:
break
x, y, z, w = env.robots[0].get_orientation()
T = np.concatenate([Quat2Rotation(-x, -y, -z, w), robot_pos.reshape(3, 1)], axis=1)
T = np.concatenate([T, np.array([[0, 0, 0, 1]])], axis=0)
# 与下一个路点的角度（用于计算下一步执行的动作）
theta = np.matmul(np.linalg.inv(T), target_pos.reshape(4, 1))
theta = theta.reshape(4)[:2]
theta /= np.linalg.norm(theta, ord=2)
theta = np.arctan2(theta[0], -theta[1]) / np.pi
# 计算与终点的角度（用于记录）
tar_theta = np.matmul(np.linalg.inv(T), np.concatenate([end_pos, np.array([1])]).reshape(4, 1))
tar_theta = tar_theta.reshape(4)[:2]
tar_theta /= np.linalg.norm(tar_theta, ord=2)
tar_theta = np.arctan2(tar_theta[0], -tar_theta[1]) / np.pi
T[:3, :3] = Quat2Rotation(x, y, z, w)
robot_T = T
# 计算下一步执行的动作
act = control(theta, state['scan'])
# 记录rgbd图
rgb = state["rgb"].transpose([2, 0, 1])
depth = state["depth"].transpose([2, 0, 1]) * 10.
observation.append(np.concatenate([depth, rgb],0))
# 记录与终点角度
target_theta.append(tar_theta)
# 记录动作
action.append(act)
# 记录位姿
robot_pos_ori.append(deepcopy(robot_T))
# 执行动作
state, reward, done, info = step(env, act, robot_T)
step_num += 1
# print(state['scan'].min()*5.6, state['scan'].argmin())
if info['collision_step'] > collison_num:
collison_num = info['collision_step']
collison_count += 1
# 若碰撞次数过多或行走步数超过500则结束本次导航任务
if collison_count > 5 or step_num > 500:
flag = True
break
# 结束本次导航任务
if flag == True:
break
# 如果完整结束导航，则保存数据
if not flag:
observation = np.stack(observation, 0).astype(np.float32)
target_theta = np.array(target_theta).astype(np.float32)
action = np.array(action).astype(np.uint8)
robot_pos_ori = np.stack(robot_pos_ori, 0).astype(np.float32)
# 保存数据
save(count, sc, observation, target_theta, action, robot_pos_ori, episode['dist'], episode['pathDist'])
count += 1
print(sc, episode['dist'], episode['pathDist'])