-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtodo.txt
172 lines (153 loc) · 8.9 KB
/
todo.txt
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
協調行動分割
協調行動が最適でなさすぎるときでも報酬を近づけてしまう
分割報酬学習
マージ
sum ある相手のある行動をとる確率×衝突したか ＝ 今までの衝突率
q学習で30％で最適解に収束しない→優先度３が正しく学習できるのは49％→優先度nが正しく学習できるかは
エキスパート特徴の時計回りと反時計回りの差が0.06は小さすぎる
逆回りのエージェントに限ってなぜか収束
"""
def collision_count(self):
for i in range(self.N_AGENTS):
collision_count = 0
for j in range(self.N_AGENTS):
if is_collision(self.coop_archive_history[i][i], self.convergence_trajs[j]):
"""
"""collision, compair num count
collision_compair[i][j] = [[trajs], [counts]]
collision_counts[i][j][1] = [collision_rate, compair_num]
selection_count[i] = [[trajs], [count]]
selection_count[i][1] = [selection_count]
"""
"""
if dose_archived:
if trajs[i] in self.selection_count[i][0]:
index = self.selection_count[i][0].index(trajs[i])
self.selection_count[i][1][index] += 1
else:
self.selection_count[i][0] += [copy.deepcopy(trajs[i])]
self.selection_count[i][1] += [1]
#print("selection_count:{}".format(self.selection_count[i]))
for j in range(self.N_AGENTS):
if i==j:
continue
traj_collision_count = 0
sum_visition_count = 0
compair_num = 0
# 衝突数の計算
for k in range(len(self.coop_archive_history[j][j])):
if self.coop_archive_history[j][j][k] in self.selection_count[j][0]:
index = self.selection_count[j][0].index(self.coop_archive_history[j][j][k])
sum_visition_count += self.selection_count[j][1][index]
else:
index = -1
sum_visition_count += 1
if is_collision(trajs[i], self.coop_archive_history[j][j][k]):
if index != -1:
traj_collision_count += self.selection_count[j][1][index]
else:
traj_collision_count += 1
if sum_visition_count!=0:
traj_collision_count /= sum_visition_count
else:
traj_collision_count /= 1
#compair_num += len(self.convergence_trajs[j])
#if compair_num==0:
if len(self.coop_archive_history[j][j])!=0:
compair_num /= len(self.coop_archive_history[j][j])
else:
compair_num = 1
#print("selectioncount:{}".format(sum_visition_count))
#print("collision_count:{}".format(traj_collision_count))
# 衝突数のカウント
if trajs[i] in self.collision_compair_memory[i][j][0]:
index = self.collision_compair_memory[i][j][0].index(trajs[i])
if not trajs[i] in self.prev_collision_compair[i][j][0]:
self.prev_collision_compair[i][j][0] += [copy.deepcopy(self.collision_compair_memory[i][j][0][index])]
self.prev_collision_compair[i][j][1] += [copy.deepcopy(self.collision_compair_memory[i][j][1][index])]
self.collision_compair_memory[i][j][1][index][0] = traj_collision_count
self.collision_compair_memory[i][j][1][index][1] = compair_num
else: # 初めてのtraj
self.prev_collision_compair[i][j][0] += [copy.deepcopy(trajs[i])]
self.prev_collision_compair[i][j][1] += [[0, 1]]
self.collision_compair_memory[i][j][0] += [copy.deepcopy(trajs[i])]
if compair_num==0:
compair_num = 1
self.collision_compair_memory[i][j][1] += [[traj_collision_count, compair_num]]
"""
"""
if dose_archived:
for j in range(self.N_AGENTS):
if i==j:
continue
traj_collision_count = 0
compair_num = 0
for k in range(len(self.coop_archive_history[j][j])):
if is_collision(trajs[i], self.coop_archive_history[j][j][k]):
traj_collision_count += 1
compair_num += len(self.coop_archive_history[j][j])
if trajs[i] in self.collision_compair_memory[i][j][0]:
index = self.collision_compair_memory[i][j][0].index(trajs[i])
if not trajs[i] in self.prev_collision_compair[i][j][0]:
self.prev_collision_compair[i][j][0] += [copy.deepcopy(self.collision_compair_memory[i][j][0][index])]
self.prev_collision_compair[i][j][1] += [copy.deepcopy(self.collision_compair_memory[i][j][1][index])]
self.collision_compair_memory[i][j][1][index][0] = self.collision_compair_memory[i][j][1][index][0] + traj_collision_count
self.collision_compair_memory[i][j][1][index][1] = self.collision_compair_memory[i][j][1][index][1] + compair_num
else:
self.prev_collision_compair[i][j][0] += [copy.deepcopy(trajs[i])]
self.prev_collision_compair[i][j][1] += [[0, 1]]
self.collision_compair_memory[i][j][0] += [copy.deepcopy(trajs[i])]
if compair_num==0:
compair_num = 1
self.collision_compair_memory[i][j][1] += [[traj_collision_count, compair_num]]
if dose_archived:
if trajs[i] in self.selection_count[i][0]:
index = self.selection_count[i][0].index(trajs[i])
self.selection_count[i][1][index] += 1
else:
self.selection_count[i][0] += [copy.deepcopy(trajs[i])]
self.selection_count[i][1] += [1]
#print("selection_count:{}".format(self.selection_count[i]))
for j in range(self.N_AGENTS):
if i==j:
continue
traj_collision_count = 0
sum_visition_count = 0
compair_num = 0
# 衝突数の計算
if self.convergence_trajs[j] in self.selection_count[j][0]:
index = self.selection_count[j][0].index(self.convergence_trajs[j])
sum_visition_count += self.selection_count[j][1][index]
else:
index = -1
sum_visition_count += 1
if is_collision(trajs[i], self.convergence_trajs[j]):
if index != -1:
traj_collision_count += self.selection_count[j][1][index]
else:
traj_collision_count += 1
if sum_visition_count!=0:
traj_collision_count /= sum_visition_count
else:
traj_collision_count /= 1
#compair_num += len(self.convergence_trajs[j])
#if compair_num==0:
compair_num=1
#print("selectioncount:{}".format(sum_visition_count))
#print("collision_count:{}".format(traj_collision_count))
# 衝突数のカウント
if trajs[i] in self.collision_compair_memory[i][j][0]:
index = self.collision_compair_memory[i][j][0].index(trajs[i])
if not trajs[i] in self.prev_collision_compair[i][j][0]:
self.prev_collision_compair[i][j][0] += [copy.deepcopy(self.collision_compair_memory[i][j][0][index])]
self.prev_collision_compair[i][j][1] += [copy.deepcopy(self.collision_compair_memory[i][j][1][index])]
self.collision_compair_memory[i][j][1][index][0] = traj_collision_count
self.collision_compair_memory[i][j][1][index][1] = compair_num
else: # 初めてのtraj
self.prev_collision_compair[i][j][0] += [copy.deepcopy(trajs[i])]
self.prev_collision_compair[i][j][1] += [[0, 1]]
self.collision_compair_memory[i][j][0] += [copy.deepcopy(trajs[i])]
if compair_num==0:
compair_num = 1
self.collision_compair_memory[i][j][1] += [[traj_collision_count, compair_num]]
"""