-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathrun_inverse_dynamics_learning.py
413 lines (319 loc) · 21.2 KB
/
run_inverse_dynamics_learning.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
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
from tensorflow.keras import layers
from typing import Any, List, Sequence, Tuple
from collections import deque, defaultdict
from absl import flags
import collections
import gym
import numpy as np
import statistics
import tensorflow as tf
import tensorflow_probability as tfp
import tqdm
import glob
import random
import retro
import gym
import os
import argparse
import cv2
import copy
import network
tfd = tfp.distributions
parser = argparse.ArgumentParser(description='Sonic Supervised Learning')
parser.add_argument('--workspace_path', type=str, help='root directory of project')
parser.add_argument('--pretrained_model', type=str, help='pretrained model name')
parser.add_argument('--replay_path', type=str, help='root directory of dataset')
parser.add_argument('--gpu_use', action='store_false', help='use gpu')
arguments = parser.parse_args()
#os.environ['CUDA_VISIBLE_DEVICES'] = '-1'
physical_devices = tf.config.list_physical_devices('GPU')
tf.config.experimental.set_memory_growth(physical_devices[0], enable=True)
workspace_path = arguments.workspace_path
#replay_path = arguments.replay_path
workspace_path = '/media/kimbring2/be356a87-def6-4be8-bad2-077951f0f3da/Sonic-the-Hedgehog-A3C-LSTM-tensorflow2'
replay_path = '/media/kimbring2/be356a87-def6-4be8-bad2-077951f0f3da/retro-movies/human/SonicTheHedgehog2-Genesis/contest/SonicTheHedgehog2-Genesis-ChemicalPlantZone.Act1'
writer = tf.summary.create_file_writer(workspace_path + "/tensorboard")
action_conversion_table = {
'[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]' : [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], # = ['']
'[0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0]' : [0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0], # = ['DOWN']
'[0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0]' : [0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0], # = ['A', 'DOWN']
'[0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]' : [0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], # = ['A']
'[0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0]' : [0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0], # = ['A', 'RIGHT']
'[0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0]' : [0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0], # = ['RIGHT']
'[0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0]' : [0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0], # = ['']
'[0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0]' : [0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0], # = ['LEFT']
'[0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0]' : [0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0], # = ['A', 'LEFT']
'[1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]' : [1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], # = ['B']
'[1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]' : [1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], # = ['B']
'[1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0]' : [1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0], # = ['B', 'RIGHT']
'[0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0]' : [0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0], # = ['LEFT']
'[0, 1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0]' : [0, 1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0], # = ['A', 'RIGHT', 'DOWN']
'[0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0]' : [1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], # = ['RIGHT']
'[0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0]' : [1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], # = ['B']
'[0, 1, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0]' : [0, 1, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0], # = ['A', 'LEFT', 'RIGHT']
'[1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0]' : [1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0], # = ['B', 'LEFT']
'[1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0]' : [1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0], # = ['B', 'DOWN']
'[1, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0]' : [1, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0], # = ['B', 'RIGHT', 'DOWN']
'[0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0]' : [0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0], # = ['LEFT']
'[1, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0]' : [1, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0], # = ['B', 'LEFT', 'RIGHT']
'[0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0]' : [0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0], # = ['UP']
'[1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0]' : [1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0], # = ['B', 'LEFT']
'[1, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0]' : [1, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0], # = ['B', 'LEFT', 'DOWN']
'[0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0]' : [0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0], # = ['DOWN', 'LEFT', 'RIGHT']
'[0, 1, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0]' : [0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0], # = ['DOWN', 'LEFT', 'RIGHT']
'[1, 1, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0]' : [0, 1, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0], # = ['A', 'LEFT', 'RIGHT']
'[1, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0]' : [1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], # = ['B']
'[0, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0]' : [1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0], # = ['B', 'RIGHT']
'[0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0, 0]' : [0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0], # = ['LEFT']
'[0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1]' : [0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0], # = ['RIGHT']
'[1, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0]' : [1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0], # = ['B', 'DOWN']
'[1, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0]' : [1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0], # = ['B', 'RIGHT’]
'[0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0]' : [1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0], # = ['B', 'LEFT']
'[1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0]' : [1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], # = ['B']
'[1, 0, 0, 0, 1, 0, 0, 1, 0, 0, 0, 0]' : [1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0], # = ['B', 'RIGHT']
'[1, 0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0]' : [1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0], # = ['B', 'RIGHT']
'[0, 0, 0, 0, 1, 0, 0, 1, 0, 0, 0, 0]' : [0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0], # = ['RIGHT']
'[0, 1, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0]' : [0, 1, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0], # = ['A', 'LEFT', 'DOWN']
'[1, 0, 0, 0, 1, 0, 1, 1, 0, 0, 0, 0]' : [1, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0], # = ['B', 'LEFT', 'RIGHT']
'[1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0]' : [1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0], # = ['B', 'RIGHT']
'[0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0]' : [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], # = ['']
'[1, 1, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0]' : [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], # = ['']
'[0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 0]' : [0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0], # = ['RIGHT']
'[1, 0, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0]' : [1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0], # = ['B', 'LEFT']
'[1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0]' : [0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0], # = ['A', 'LEFT']
'[1, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0, 0]' : [1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], # = ['B']
'[0, 0, 0, 0, 1, 0, 1, 1, 0, 0, 0, 0]' : [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], # = ['']
'[0, 1, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0]' : [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], # = ['']
'[1, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0]' : [1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], # = ['B']
'[1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 1]' : [0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0], # = ['LEFT']
'[0, 0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0]' : [1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0], # = ['B', 'RIGHT']
'[1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1]' : [1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0], # = ['B', 'RIGHT']
'[1, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 0]' : [1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0], # = ['B', 'RIGHT']
'[0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0]' : [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], # = ['']
'[1, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 1]' : [0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0], # = ['RIGHT']
'[1, 0, 0, 0, 1, 0, 0, 1, 0, 1, 0, 0]' : [1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0], # = ['B', 'RIGHT']
'[1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0]' : [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], # = ['']
'[0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 1, 0]' : [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], # = ['']
'[1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 1]' : [1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0], # = ['B', 'LEFT']
'[1, 0, 0, 0, 0, 1, 0, 0, 1, 0, 0, 0]' : [1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0], # = ['B', 'DOWN']
'[0, 1, 0, 0, 0, 0, 1, 1, 0, 0, 1, 0]' : [1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], # = ['B']
'[0, 1, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0]' : [0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0], # = ['RIGHT']
'[0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0]' : [1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], # = ['B']
'[1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0]' : [1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], # = ['B']
'[0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0]' : [1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0], # = ['B', 'RIGHT']
'[0, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0]' : [0, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0], # = ['DOWN', RIGHT']
'[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1]' : [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0] # = ['']
}
# ['']: [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
# ['LEFT']: [0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0]
# ['RIGHT']: [0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0]
# ['B']: [1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
# ['DOWN']: [0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0]
# ['A']: [0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
# ['UP']: [0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0]
# ['B', 'LEFT']: [1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0]
# ['B', 'RIGHT']: [1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0]
# ['B', 'DOWN']: [1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0]
# ['A', 'LEFT']: [0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0]
# ['A', 'RIGHT']: [0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0]
# ['A', 'DOWN']: [0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0]
# ['DOWN', 'RIGHT']: [0, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0]
# ['DOWN', 'LEFT']: [0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0]
# ['LEFT', 'RIGHT']: [0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0]
# ['B', 'RIGHT', 'DOWN']: [1, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0]
# ['B', 'LEFT', 'DOWN'] : [1, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0]
# ['A', 'LEFT', 'DOWN'] : [0, 1, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0]
# ['A', 'LEFT', 'RIGHT']: [0, 1, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0]
# ['A', 'RIGHT', 'DOWN']: [0, 1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0]
# ['B', 'LEFT', 'RIGHT']: [1, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0]
# ['DOWN', 'LEFT', 'RIGHT']: [0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0]
possible_action_list = [
[0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0],
[1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0],
[0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0],
[1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0],
[1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0],
[1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0],
[0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0],
[0, 1, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0],
[0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0],
[1, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0],
[1, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0],
[0, 1, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0],
[0, 1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0],
[1, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 1, 1, 1, 0, 0, 0, 0],
[0, 1, 0, 0, 0, 1, 1, 0, 0, 0, 0, 0]
]
#print("len(possible_action_list): ", len(possible_action_list))
stage_name_list = ['EmeraldHillZone', 'ChemicalPlantZone', 'AquaticRuinZone', 'CasinoNightZone', 'HillTopZone',
'MysticCaveZone', 'OilOceanZone', 'MetropolisZone', 'WingFortressZone']
stage_len = len(stage_name_list)
env = retro.make(game='SonicTheHedgehog2-Genesis', state=retro.State.NONE, use_restricted_actions=retro.Actions.ALL)
def one_hot(a, num_classes):
return np.squeeze(np.eye(num_classes)[a])
time_step = 64
class TrajetoryDataset(tf.data.Dataset):
def _generator(num_trajectorys):
replay_file_path_list = glob.glob(replay_path + '/*.bk2')
#while True:
for replay_name in replay_file_path_list:
#replay_file_path_list = glob.glob(replay_path + '/*.bk2')
#replay_name = random.choice(replay_file_path_list)
replay_name = replay_name.split('/')[-1]
replay = retro.Movie(os.path.join(replay_path, replay_name))
replay.step()
env.initial_state = replay.get_state()
obs = env.reset()
obs = cv2.resize(obs, dsize=(84, 84), interpolation=cv2.INTER_AREA)
obs = obs / 255.0
action_index = 0
pre_action_index = action_index
obs_history_list, action_history_list, action_list, raw_action_list = [], [], [], []
obs_history_list_list, action_history_list_list, action_list_list, raw_action_list_list = [], [], [], []
action_history = np.zeros((12, len(possible_action_list)))
step_num = 0
print('stepping replay')
offset_step = random.randint(0, 64)
while replay.step():
#env.render()
#print("step_num: ", step_num)
keys = []
for i in range(len(env.buttons)):
key = int(replay.get_key(i, 0))
keys.append(key)
converted_action = action_conversion_table[str(keys)]
pre_action_index = action_index
action_index = possible_action_list.index(converted_action)
next_obs, rew, done, info = env.step(converted_action)
next_obs = cv2.resize(next_obs, dsize=(84, 84), interpolation=cv2.INTER_AREA)
next_obs = next_obs / 255.0
########################################################################################
if offset_step == 0:
obs_history_list.append(obs)
raw_action_list.append(action_index)
action_onehot = one_hot(action_index, num_actions)
action_list.append(action_onehot)
pre_action_onehot = one_hot(pre_action_index, num_actions)
action_history = np.roll(action_history, 1, axis=0)
action_history[0,:] = pre_action_onehot
action_history_list.append(action_history)
else:
action_onehot = one_hot(action_index, num_actions)
pre_action_onehot = one_hot(pre_action_index, num_actions)
action_history = np.roll(action_history, 1, axis=0)
action_history[0,:] = pre_action_onehot
offset_step -= 1
########################################################################################
obs = next_obs
saved_state = env.em.get_state()
if len(obs_history_list) == time_step:
obs_history_list_list.append(obs_history_list)
action_history_list_list.append(action_history_list)
action_list_list.append(action_list)
raw_action_list_list.append(raw_action_list)
obs_history_list, action_history_list, action_list, raw_action_list = [], [], [], []
step_num += 1
yield (obs_history_list_list, action_history_list_list, action_list_list, raw_action_list_list)
def __new__(cls, num_trajectorys=3):
return tf.data.Dataset.from_generator(cls._generator, output_types=(tf.dtypes.float32, tf.dtypes.float32, tf.dtypes.float32,
tf.dtypes.float32), args=(num_trajectorys,))
dataset = tf.data.Dataset.range(1).interleave(TrajetoryDataset,
num_parallel_calls=tf.data.experimental.AUTOTUNE).batch(1).prefetch(tf.data.experimental.AUTOTUNE)
num_actions = len(possible_action_list)
num_hidden_units = 1024
model = network.InverseActionPolicy(num_actions, num_hidden_units)
model_name = 'inverse_dynamic_model_20'
#model.load_weights("model/" + model_name)
if arguments.pretrained_model != None:
print("Load Pretrained Model")
model.load_weights("model/" + arguments.pretrained_model)
cce_loss = tf.keras.losses.CategoricalCrossentropy()
cce_loss_logits = tf.keras.losses.CategoricalCrossentropy(from_logits=True)
optimizer = tf.keras.optimizers.Adam(0.00001)
def supervised_replay(obs_list, act_list, act_history_list, raw_act_list, memory_state_obs, carry_state_obs, memory_state_his, carry_state_his):
memory_state_obs = tf.concat(memory_state_obs, 0)
carry_state_obs = tf.concat(carry_state_obs, 0)
memory_state_his = tf.concat(memory_state_his, 0)
carry_state_his = tf.concat(carry_state_his, 0)
#print("obs_list.shape: ", obs_list.shape)
#print("act_list.shape: ", act_list.shape)
#print("act_history_list.shape: ", act_history_list.shape)
with tf.GradientTape() as tape:
prediction = model(obs_list, act_history_list, memory_state_obs, carry_state_obs,
memory_state_his, carry_state_his, training=True)
act_pi = prediction[0]
memory_state_obs = prediction[1]
carry_state_obs = prediction[2]
memory_state_his = prediction[3]
carry_state_his = prediction[4]
#print("act_list[0].shape: ", act_list[0].shape)
#print("act_pi[0].shape: ", act_pi[0].shape)
action_dist = tfd.Categorical(logits=act_pi)
action_indexs = action_dist.sample()
#print("raw_act_list: ", raw_act_list)
#print("act_list: ", act_list)
#print("action_indexs: ", action_indexs)
act_loss = cce_loss_logits(act_list[0], act_pi[0])
#print("act_loss: ", act_loss)
#print("")
regularization_loss = tf.reduce_sum(model.losses)
total_loss = act_loss + 1e-5 * regularization_loss
grads = tape.gradient(total_loss, model.trainable_variables)
optimizer.apply_gradients(zip(grads, model.trainable_variables))
return total_loss, memory_state_obs, carry_state_obs, memory_state_his, carry_state_his
def supervised_train(dataset):
training_episode = 0
for batch in dataset:
episode_size = batch[0].shape[1]
print("episode_size: ", episode_size)
replay_obs_array = batch[0][0]
replay_act_his_array = batch[1][0]
replay_act_array = batch[2][0]
replay_raw_act_array = batch[3][0]
#print("replay_obs_array.shape: ", replay_obs_array.shape)
#print("replay_act_his_array.shape: ", replay_act_his_array.shape)
#print("replay_act_array.shape: ", replay_act_array.shape)
#print("replay_raw_act_array.shape: ", replay_raw_act_array.shape)
memory_state_obs = np.zeros([1,512], dtype=np.float32)
carry_state_obs = np.zeros([1,512], dtype=np.float32)
memory_state_his = np.zeros([1,128], dtype=np.float32)
carry_state_his = np.zeros([1,128], dtype=np.float32)
step_length = 1
batch_size = replay_obs_array.shape[0]
train_losses = []
for episode_index in range(0, episode_size, step_length):
#print("episode_index: ", episode_index)
obs = replay_obs_array[episode_index:episode_index+step_length,:,:,:,:]
act_history = replay_act_his_array[episode_index:episode_index+step_length,:,:,:]
act = replay_act_array[episode_index:episode_index+step_length,:]
raw_act = replay_raw_act_array[episode_index:episode_index+step_length,:]
#print("episode_index: ", episode_index)
if len(obs) != step_length:
break
total_loss, memory_state_obs, carry_state_obs, memory_state_his, carry_state_his = supervised_replay(obs, act, act_history,
raw_act,
memory_state_obs,
carry_state_obs,
memory_state_his,
carry_state_his)
train_losses.append(total_loss)
#print("total_loss: ", total_loss)
#print("")
mean_loss_train = np.mean(train_losses)
print("Mean train loss:", mean_loss_train)
print("")
with writer.as_default():
tf.summary.scalar("mean_loss_train", mean_loss_train, step=training_episode)
writer.flush()
if training_episode % 10 == 0:
model.save_weights(workspace_path + '/model/inverse_dynamic_model_' + str(training_episode))
training_episode += 1
supervised_train(dataset)