forked from hewudi666/Graph_RL
-
Notifications
You must be signed in to change notification settings - Fork 0
/
runner_ppo_lstm.py
281 lines (259 loc) · 10.9 KB
/
runner_ppo_lstm.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
import torch
import os
import numpy as np
import logging
from collections import namedtuple
import time
import torch
import torch.nn as nn
import torch.nn.functional as F
import matplotlib.pyplot as plt
Transition = namedtuple('Transition', ['state', 'action', 'a_log_prob', 'reward', 'next_state'])
class embedding(nn.Module):
def __init__(self, args):
super(embedding, self).__init__()
intruder_num = args.n_agents - 1
self.lstm = nn.LSTM(intruder_num * 9, 32)
self.fc1 = nn.Linear(32 + 9, 256)
self.fc2 = nn.Linear(256, 256)
self.fc3 = nn.Linear(256, 32)
def forward(self, obs, obs_in):
"""
:param obs: (9, )
:param obs_in: ((n-1)*9 ,1)
:return:
"""
obs_in = obs_in.view(-1, 1, len(obs_in))
x, _ = self.lstm(obs_in)
x = x.squeeze(dim=1)
obs = obs.view(-1, len(obs))
x = torch.cat([x, obs], dim=1)
x = F.relu(self.fc1(x))
x = F.relu(self.fc2(x))
x = F.relu(self.fc3(x))
x = x.view(x.size()[0], -1) # torch size [1, 32]
return x
class Runner_PPO_LSTM:
def __init__(self, args, env):
self.args = args
self.epsilon = args.epsilon
self.max_step = args.max_episode_len
self.env = env
self.agents = self.env.agents
self.agent_num = self.env.agent_num
self.lstm = embedding(args)
self.save_path = self.args.save_dir + '/' + self.args.scenario_name
if not os.path.exists(self.save_path):
os.makedirs(self.save_path)
def get_state(self, obs):
"""
:param obs: numpy [agent_n, 9]
:return: numpy [agent_n, 32]
"""
state = []
for i, agent in enumerate(self.agents):
obs_own = obs[i] # (9, )
obs_intruder = np.delete(obs, i, axis=0).reshape(-1, 1)
encoder = self.lstm(torch.Tensor(obs_own), torch.Tensor(obs_intruder))
ob = encoder.detach().numpy()[0] # (32, )
state.append(ob)
return np.array(state)
def run(self):
returns = []
reward_total = []
conflict_total = []
collide_wall_total = []
nmac_total = []
success_total = []
start = time.time()
for episode in range(self.args.num_episodes):
reward_episode = []
steps = 0
self.epsilon = max(0.05, self.epsilon - 0.0004)
s = self.env.reset()
s = self.get_state(s)
print("current_episode {}".format(episode))
while steps < self.max_step:
if not self.env.simulation_done:
actions = []
action_probs = []
for i, agent in enumerate(self.agents):
action, action_prob = agent.policy.select_action(s[i])
actions.append(action)
action_probs.append(action_prob)
s_next, r, done, info = self.env.step(actions)
s_next = self.get_state(s_next)
for i, agent in enumerate(self.agents):
trans = Transition(s[i], actions[i], action_probs[i], r[i], s_next[i])
agent.policy.store_transition(trans)
s = s_next
reward_episode.append(sum(r) / 1000)
else:
# print("robot_terminated_times:", self.env.agent_times)
if self.env.simulation_done:
print("all agent done!")
for i, agent in enumerate(self.agents):
if len(agent.policy.buffer) >= self.args.batch_size:
agent.policy.update()
break
reward_total.append(sum(reward_episode))
if episode > 0 and episode % self.args.evaluate_rate == 0:
rew, info = self.evaluate()
if episode % (5 * self.args.evaluate_rate) == 0:
self.env.render(mode='traj')
returns.append(rew)
conflict_total.append(info[0])
collide_wall_total.append(info[1])
success_total.append(info[2])
nmac_total.append(info[3])
self.env.conflict_num_episode = 0
self.env.nmac_num_episode = 0
end = time.time()
print("花费时间", end - start)
plt.figure()
plt.plot(range(1, len(returns)), returns[1:])
plt.xlabel('evaluate num')
plt.ylabel('average returns')
plt.savefig(self.save_path + '/15_train_return.png', format='png')
np.save(self.save_path + '/15_train_returns', returns)
fig, a = plt.subplots(2, 2)
x = range(len(conflict_total))
a[0][0].plot(x, conflict_total, 'b')
a[0][0].set_title('conflict_num')
a[0][1].plot(x, collide_wall_total, 'y')
a[0][1].set_title('exit_boundary_num')
a[1][0].plot(x, success_total, 'r')
a[1][0].set_title('success_num')
a[1][1].plot(x, nmac_total)
a[1][1].set_title('nmac_num')
plt.savefig(self.save_path + '/15_train_metric.png', format='png')
np.save(self.save_path + '/15_train_conflict', conflict_total)
plt.show()
def evaluate(self):
print("now is evaluate!")
self.env.collision_num = 0
self.env.exit_boundary_num = 0
self.env.success_num = 0
self.env.nmac_num = 0
returns = []
deviation = []
for episode in range(self.args.evaluate_episodes):
# reset the environment
s = self.env.reset()
s = self.get_state(s)
rewards = 0
for time_step in range(self.args.evaluate_episode_len):
# self.env.render()
if not self.env.simulation_done:
actions = []
for agent_id, agent in enumerate(self.agents):
action, action_prob = agent.policy.select_action(s[agent_id])
actions.append(action)
s_next, r, done, info = self.env.step(actions)
s_next = self.get_state(s_next)
rewards += sum(r)
s = s_next
else:
dev = self.env.route_deviation_rate()
deviation.append(np.mean(dev))
break
rewards = rewards / 10000
returns.append(rewards)
print('Returns is', rewards)
print("conflict num :", self.env.collision_num)
print("nmac num:", self.env.nmac_num)
print("exit boundary num:", self.env.exit_boundary_num)
print("success num:", self.env.success_num)
print("路径平均偏差率:", np.mean(deviation))
return sum(returns) / self.args.evaluate_episodes, (self.env.collision_num, self.env.exit_boundary_num, self.env.success_num, self.env.nmac_num)
def evaluate_model(self):
"""
对现有最新模型进行评估
:return:
"""
print("now evaluate the ppo model")
conflict_total = []
collide_wall_total = []
success_total = []
nmac_total = []
self.env.collision_num = 0
self.env.nmac_num = 0
self.env.exit_boundary_num = 0
self.env.success_num = 0
returns = []
deviation = []
eval_episode = 100
for episode in range(eval_episode):
# reset the environment
s = self.env.reset()
rewards = 0
s = self.get_state(s)
for time_step in range(self.args.evaluate_episode_len):
# self.env.render()
if not self.env.simulation_done:
actions = []
for agent_id, agent in enumerate(self.agents):
action, action_prob = agent.policy.select_action(s[agent_id])
actions.append(action)
s_next, r, done, info = self.env.step(actions)
s_next = self.get_state(s_next)
rewards += sum(r)
s = s_next
else:
dev = self.env.route_deviation_rate()
deviation.append(np.mean(dev))
break
if episode > 0 and episode % 50 == 0:
self.env.render(mode='traj')
# plt.figure()
# plt.title('collision_value——time')
# x = range(len(self.env.collision_value))
# plt.plot(x, self.env.collision_value)
# plt.xlabel('timestep')
# plt.ylabel('collision_value')
# plt.savefig(self.save_path + '/collision_value/30_agent/' + str(episode) + 'collision_value.png', format='png')
# np.save(self.save_path + '/collision_value/30_agent/' + str(episode) + 'collision_value.npy', self.env.collision_value)
# plt.close()
rewards = rewards / 1000
returns.append(rewards)
print('Returns is', rewards)
print("conflict num :", self.env.collision_num)
print("nmac num:", self.env.nmac_num)
print("exit boundary num:", self.env.exit_boundary_num)
print("success num:", self.env.success_num)
conflict_total.append(self.env.collision_num)
nmac_total.append(self.env.nmac_num)
collide_wall_total.append(self.env.exit_boundary_num)
success_total.append(self.env.success_num)
self.env.collision_num = 0
self.env.nmac_num = 0
self.env.exit_boundary_num = 0
self.env.success_num = 0
plt.figure()
plt.plot(range(1, len(returns)), returns[1:])
plt.xlabel('evaluate num')
plt.ylabel('average returns')
# plt.savefig(self.save_path + '/15_eval_return.png', format='png')
fig, a = plt.subplots(2, 2)
x = range(len(conflict_total))
ave_conflict = np.mean(conflict_total)
ave_nmac = np.mean(nmac_total)
ave_success = np.mean(success_total)
ave_exit = np.mean(collide_wall_total)
zero_conflict = sum(np.array(conflict_total) == 0)
print("平均冲突数", ave_conflict)
print("平均NMAC数", ave_nmac)
print("平均成功率", ave_success / self.agent_num)
print("平均出界率", ave_exit / self.agent_num)
print("0冲突占比:", zero_conflict / len(conflict_total))
print("平均偏差率", np.mean(deviation))
a[0][0].plot(x, conflict_total, 'b')
a[0][0].set_title('conflict_num')
a[0][1].plot(x, collide_wall_total, 'y')
a[0][1].set_title('exit_boundary_num')
a[1][0].plot(x, success_total, 'r')
a[1][0].set_title('success_num')
a[1][1].plot(x, nmac_total)
a[1][1].set_title('nmac_num')
# plt.savefig(self.save_path + '/15_eval_metric.png', format='png')
plt.show()