forked from Acmece/rl-collision-avoidance
-
Notifications
You must be signed in to change notification settings - Fork 1
/
circle_world.py
280 lines (216 loc) · 9 KB
/
circle_world.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
import time
import rospy
import copy
import tf
import numpy as np
from geometry_msgs.msg import Twist, Pose
from nav_msgs.msg import Odometry
from sensor_msgs.msg import LaserScan
from rosgraph_msgs.msg import Clock
from std_srvs.srv import Empty
from std_msgs.msg import Int8
from model.utils import test_init_pose, test_goal_point
class StageWorld():
def __init__(self, beam_num, index, num_env):
self.index = index
self.num_env = num_env
node_name = 'StageEnv_' + str(index)
rospy.init_node(node_name, anonymous=None)
self.beam_mum = beam_num
self.laser_cb_num = 0
self.scan = None
# used in reset_world
self.self_speed = [0.0, 0.0]
self.step_goal = [0., 0.]
self.step_r_cnt = 0.
# used in generate goal point
self.map_size = np.array([8., 8.], dtype=np.float32) # 20x20m
self.goal_size = 0.5
self.robot_value = 10.
self.goal_value = 0.
# -----------Publisher and Subscriber-------------
cmd_vel_topic = 'robot_' + str(index) + '/cmd_vel'
self.cmd_vel = rospy.Publisher(cmd_vel_topic, Twist, queue_size=10)
cmd_pose_topic = 'robot_' + str(index) + '/cmd_pose'
self.cmd_pose = rospy.Publisher(cmd_pose_topic, Pose, queue_size=10)
object_state_topic = 'robot_' + str(index) + '/base_pose_ground_truth'
self.object_state_sub = rospy.Subscriber(object_state_topic, Odometry, self.ground_truth_callback)
laser_topic = 'robot_' + str(index) + '/base_scan'
self.laser_sub = rospy.Subscriber(laser_topic, LaserScan, self.laser_scan_callback)
odom_topic = 'robot_' + str(index) + '/odom'
self.odom_sub = rospy.Subscriber(odom_topic, Odometry, self.odometry_callback)
crash_topic = 'robot_' + str(index) + '/is_crashed'
self.check_crash = rospy.Subscriber(crash_topic, Int8, self.crash_callback)
self.sim_clock = rospy.Subscriber('clock', Clock, self.sim_clock_callback)
# -----------Service-------------------
self.reset_stage = rospy.ServiceProxy('reset_positions', Empty)
# # Wait until the first callback
self.speed = None
self.state = None
self.speed_GT = None
self.state_GT = None
self.is_crashed = None
while self.scan is None or self.speed is None or self.state is None\
or self.speed_GT is None or self.state_GT is None or self.is_crashed is None:
pass
rospy.sleep(1.)
# # What function to call when you ctrl + c
# rospy.on_shutdown(self.shutdown)
def ground_truth_callback(self, GT_odometry):
Quaternious = GT_odometry.pose.pose.orientation
Euler = tf.transformations.euler_from_quaternion([Quaternious.x, Quaternious.y, Quaternious.z, Quaternious.w])
self.state_GT = [GT_odometry.pose.pose.position.x, GT_odometry.pose.pose.position.y, Euler[2]]
v_x = GT_odometry.twist.twist.linear.x
v_y = GT_odometry.twist.twist.linear.y
v = np.sqrt(v_x**2 + v_y**2)
self.speed_GT = [v, GT_odometry.twist.twist.angular.z]
def laser_scan_callback(self, scan):
self.scan_param = [scan.angle_min, scan.angle_max, scan.angle_increment, scan.time_increment,
scan.scan_time, scan.range_min, scan.range_max]
self.scan = np.array(scan.ranges)
self.laser_cb_num += 1
def odometry_callback(self, odometry):
Quaternions = odometry.pose.pose.orientation
Euler = tf.transformations.euler_from_quaternion([Quaternions.x, Quaternions.y, Quaternions.z, Quaternions.w])
self.state = [odometry.pose.pose.position.x, odometry.pose.pose.position.y, Euler[2]]
self.speed = [odometry.twist.twist.linear.x, odometry.twist.twist.angular.z]
def sim_clock_callback(self, clock):
self.sim_time = clock.clock.secs + clock.clock.nsecs / 1000000000.
def crash_callback(self, flag):
self.is_crashed = flag.data
def get_self_stateGT(self):
return self.state_GT
def get_self_speedGT(self):
return self.speed_GT
def get_laser_observation(self):
scan = copy.deepcopy(self.scan)
scan[np.isnan(scan)] = 6.0
scan[np.isinf(scan)] = 6.0
raw_beam_num = len(scan)
sparse_beam_num = self.beam_mum
step = float(raw_beam_num) / sparse_beam_num
sparse_scan_left = []
index = 0.
for x in xrange(int(sparse_beam_num / 2)):
sparse_scan_left.append(scan[int(index)])
index += step
sparse_scan_right = []
index = raw_beam_num - 1.
for x in xrange(int(sparse_beam_num / 2)):
sparse_scan_right.append(scan[int(index)])
index -= step
scan_sparse = np.concatenate((sparse_scan_left, sparse_scan_right[::-1]), axis=0)
return scan_sparse / 6.0 - 0.5
def get_self_speed(self):
return self.speed
def get_self_state(self):
return self.state
def get_crash_state(self):
return self.is_crashed
def get_sim_time(self):
return self.sim_time
def get_local_goal(self):
[x, y, theta] = self.get_self_stateGT()
[goal_x, goal_y] = self.goal_point
local_x = (goal_x - x) * np.cos(theta) + (goal_y - y) * np.sin(theta)
local_y = -(goal_x - x) * np.sin(theta) + (goal_y - y) * np.cos(theta)
return [local_x, local_y]
def reset_world(self):
self.reset_stage()
self.self_speed = [0.0, 0.0]
self.step_goal = [0., 0.]
self.step_r_cnt = 0.
self.start_time = time.time()
rospy.sleep(0.5)
def generate_goal_point(self):
self.goal_point = test_goal_point(self.index)
self.pre_distance = 0
self.distance = copy.deepcopy(self.pre_distance)
def get_reward_and_terminate(self, t):
terminate = False
laser_scan = self.get_laser_observation()
[x, y, theta] = self.get_self_stateGT()
[v, w] = self.get_self_speedGT()
self.pre_distance = copy.deepcopy(self.distance)
self.distance = np.sqrt((self.goal_point[0] - x) ** 2 + (self.goal_point[1] - y) ** 2)
reward_g = (self.pre_distance - self.distance) * 2.5
reward_c = 0
reward_w = 0
result = 0
is_crash = self.get_crash_state()
if self.distance < self.goal_size:
terminate = True
reward_g = 15
result = 'Reach Goal'
if is_crash == 1:
terminate = True
reward_c = -15.
result = 'Crashed'
if np.abs(w) > 0.7:
reward_w = -0.1 * np.abs(w)
if t > 10000:
terminate = True
result = 'Time out'
reward = reward_g + reward_c + reward_w
return reward, terminate, result
def reset_pose(self):
reset_pose = test_init_pose(self.index)
self.control_pose(reset_pose)
def control_vel(self, action):
move_cmd = Twist()
move_cmd.linear.x = action[0]
move_cmd.linear.y = 0.
move_cmd.linear.z = 0.
move_cmd.angular.x = 0.
move_cmd.angular.y = 0.
move_cmd.angular.z = action[1]
self.cmd_vel.publish(move_cmd)
def control_pose(self, pose):
pose_cmd = Pose()
assert len(pose)==3
pose_cmd.position.x = pose[0]
pose_cmd.position.y = pose[1]
pose_cmd.position.z = 0
qtn = tf.transformations.quaternion_from_euler(0, 0, pose[2], 'rxyz')
pose_cmd.orientation.x = qtn[0]
pose_cmd.orientation.y = qtn[1]
pose_cmd.orientation.z = qtn[2]
pose_cmd.orientation.w = qtn[3]
self.cmd_pose.publish(pose_cmd)
def generate_random_pose(self):
[x_robot, y_robot, theta] = self.get_self_stateGT()
x = np.random.uniform(9, 19)
y = np.random.uniform(0, 1)
if y <= 0.4:
y = -(y * 10 + 1)
else:
y = -(y * 10 + 9)
dis_goal = np.sqrt((x - x_robot) ** 2 + (y - y_robot) ** 2)
while (dis_goal < 7) and not rospy.is_shutdown():
x = np.random.uniform(9, 19)
y = np.random.uniform(0, 1)
if y <= 0.4:
y = -(y * 10 + 1)
else:
y = -(y * 10 + 9)
dis_goal = np.sqrt((x - x_robot) ** 2 + (y - y_robot) ** 2)
theta = np.random.uniform(0, 2*np.pi)
return [x, y, theta]
def generate_random_goal(self):
[x_robot, y_robot, theta] = self.get_self_stateGT()
x = np.random.uniform(9, 19)
y = np.random.uniform(0, 1)
if y <= 0.4:
y = -(y*10 + 1)
else:
y = -(y*10 + 9)
dis_goal = np.sqrt((x - x_robot) ** 2 + (y - y_robot) ** 2)
while (dis_goal < 7) and not rospy.is_shutdown():
x = np.random.uniform(9, 19)
y = np.random.uniform(0, 1)
if y <= 0.4:
y = -(y * 10 + 1)
else:
y = -(y * 10 + 9)
dis_goal = np.sqrt((x - x_robot) ** 2 + (y - y_robot) ** 2)
return [x, y]