/
games_wrapper.py
480 lines (398 loc) · 15 KB
/
games_wrapper.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
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
#from vizdoom import *
import scipy
import numpy as np
from ple.games.catcher import Catcher
from ple.games.raycastmaze import RaycastMaze
from ple.games.pixelcopter_v2 import Pixelcopter_v2
from ple import PLE
from PIL import Image
import deepmind_lab
import random
import cv2
seed = 147
random.seed(seed)
import os
import sys
unbuffered = os.fdopen(sys.stdout.fileno(), 'w', 0)
sys.stdout = unbuffered
class LabWrapper:
def __init__(self, width):
self.game = None
self.actions = None
self.width = width
self.frames_played = 0
self.max_game_len = 3000
self.visitation_map = {}
self.reward_positions = []
# Create game env
self.game = self.set_lab_game_setup()
# Reset game
self.restart_game()
self.top_down_view=None
def construct_visitation_map(self):
return None
def _action(self,*entries):
return np.array(entries, dtype=np.intc)
def set_lab_game_setup(self):
level = 'nav_maze_static_01'
level = 'small_maze'
#level = 'small_maze_multimap'
env = deepmind_lab.Lab(
level,
['RGB_INTERLACED',
'DEBUG.POS.TRANS',
'DEBUG.CAMERA.TOP_DOWN'],
config={
'fps': str(60),
'width': str(self.width),
'height': str(self.width)
})
self.actions = [
self._action(-20, 0, 0, 0, 0, 0, 0), # look_left
self._action(20, 0, 0, 0, 0, 0, 0), # look_right
# _action( 0, 10, 0, 0, 0, 0, 0), # look_up
# _action( 0, -10, 0, 0, 0, 0, 0), # look_down
self._action(0, 0, -1, 0, 0, 0, 0), # strafe_left
self._action(0, 0, 1, 0, 0, 0, 0), # strafe_right
self._action(0, 0, 0, 1, 0, 0, 0), # forward
self._action(0, 0, 0, -1, 0, 0, 0), # backward
# _action( 0, 0, 0, 0, 1, 0, 0), # fire
# _action( 0, 0, 0, 0, 0, 1, 0), # jump
# _action( 0, 0, 0, 0, 0, 0, 1) # crouch
]
return env
def restart_game(self):
self.game.reset()
self.frames_played = 0
self.visitation_map[self.frames_played] = self.game.observations()[
'DEBUG.POS.TRANS']
#the starting point is random through env
def process_frame(self, image):
image = image.astype(np.float32)
#normalize
image = np.reshape(image, [np.prod(image.shape), 1]) / 255.0
return image
def get_frame(self):
colour_frame = self.game.observations()['RGB_INTERLACED']
frame = cv2.cvtColor( colour_frame, cv2.COLOR_RGB2GRAY )
#cv2.imshow('Image',self.last_frame)
#cv2.waitKey(0)
#cv2.destroyAllWindows()
return frame, colour_frame
def game_finished(self):
if self.frames_played >= self.max_game_len:
print('LOG: max frames')
return True
if not self.game.is_running():
print('LOG: game ended from engine')
return not self.game.is_running()
def make_action(self, action_index):
#print ('ACTION index %d action %s'%(action_index,str(self.actions[action_index])))
reward = self.game.step(self.actions[action_index], num_steps=4)
self.frames_played += 1
if self.game_finished():
self.visitation_map[self.frames_played] = self.visitation_map[
self.frames_played - 1] + (1,0,0)
else:
self.visitation_map[self.frames_played] = self.game.observations()['DEBUG.POS.TRANS']
if reward > 0 and not self.game_finished():
self.reward_positions.append(self.game.observations()[
'DEBUG.POS.TRANS'])
return reward
def construct_visitation_map(self):
# mark rewards on map with red
max_pos_x = 0
max_pos_y = 0
min_pos_x = 9999
min_pos_y = 9999
for position in self.visitation_map:
(x,y,rot) = self.visitation_map[position]
if x > max_pos_x:
max_pos_x = x
if x < min_pos_x:
min_pos_x = x
if y > max_pos_y:
max_pos_y = y
if y < min_pos_y:
min_pos_y = y
print('image of shape (%d,%d,%d)'%(483,483,4))
print('bias x %d bias y %d'%(min_pos_x,min_pos_y))
print('max pos x %d max pos y %d'%(max_pos_x, max_pos_y))
image = np.ones((int(483+116), int(483+116),4))
image.fill(255)
image[:,:,3].fill(0)
transparency = np.linspace(20,255,len(self.visitation_map))
print('total steps %d '%len(self.visitation_map))
step_color = (128,0,255)
for timestep in range(len(self.visitation_map)):
(x,y,z) = self.visitation_map[timestep]
x = int(x)
y = int(y)
if timestep == 0 :
#red is start point
cv2.circle(image, (x,y), 7, (0,0,255,255), thickness=-1,
lineType=8, shift=0)
elif timestep == len(self.visitation_map) - 1:
# blue is end point
cv2.circle(image, (x, y), 7, (255, 0 , 0,255), thickness=-1,
lineType=8, shift=0)
else:
cv2.circle(image, (x, y), 4, (128,0,128,int(transparency[
timestep])),
thickness=-1,
lineType=8, shift=0)
return image
''' Doom game class'''
class DoomWrapper:
def __init__(self, width):
'''
@width : width of game window
'''
self.game = None
self.max_game_len = 300
self.actions = [[True, False, False], [False, True, False], [False, False, True]]
self.width = width
#Create game env
self.game = self.set_doom_game_setup(self.max_game_len)
def set_doom_game_setup(self, max_game_len):
'''
@max_game_len : maximum time steps allowed before terminating episode
'''
game = DoomGame()
game.set_doom_scenario_path("basic.wad") # This corresponds to the simple task we will pose our agent
game.set_doom_map("map01")
game.set_screen_resolution(ScreenResolution.RES_160X120)
game.set_screen_format(ScreenFormat.GRAY8)
game.set_render_hud(False)
game.set_render_crosshair(False)
game.set_render_weapon(True)
game.set_render_decals(False)
game.set_render_particles(False)
game.add_available_button(Button.MOVE_LEFT)
game.add_available_button(Button.MOVE_RIGHT)
game.add_available_button(Button.ATTACK)
game.add_available_game_variable(GameVariable.AMMO2)
game.add_available_game_variable(GameVariable.POSITION_X)
game.add_available_game_variable(GameVariable.POSITION_Y)
game.set_episode_timeout(max_game_len)
game.set_episode_start_time(10)
game.set_window_visible(False)
game.set_sound_enabled(False)
game.set_living_reward(-1)
game.set_mode(Mode.PLAYER)
game.init()
return game
def restart_game(self):
self.game.new_episode()
def get_frame(self):
frame = self.game.get_state().screen_buffer
return frame
''' Processes Doom screen image to produce cropped and resized image.'''
def process_frame(self, frame):
processed = frame[10:-10, 30:-30]
processed = scipy.misc.imresize(processed, [self.width, self.width])
# also normalize
processed = np.reshape(processed, [np.prod(processed.shape)]) / 255.0
return processed
def game_finished(self):
return self.game.is_episode_finished()
def make_action(self, action_index):
reward = self.env.make_action(self.actions[action_index]) / 100.0
return reward
''' PLE Catcher game '''
class CatcherWrapper:
def __init__(self, width, lives = 1):
'''
@width : width of game window
@lives : number of deaths before the episode terminates (death = pallet does not catch ball)
'''
self.width = width
self.game = None
self.actions = None
self.max_game_len = 150
self.visitation_map = {}
self.timer = 0
self.coordinates = (0, 0)
# Create game env
catcher = Catcher(width=width, height=width,init_lives=lives)
self.game = self.set_catcher_game_setup(catcher)
def set_catcher_game_setup(self, game):
p = PLE(game, display_screen=False)
self.actions = p.getActionSet()
p.init()
return p
def restart_game(self):
self.visitation_map = {}
self.timer = 0
self.coordinates = (0, 0)
self.game.reset_game()
frame_skip = random.randint(0,30)
#Randomize start
for i in range(frame_skip):
reward = self.make_action(random.choice(range(len(self.actions))))
self.coordinates = (self.game.game.getGameState()['player_x'],
10)
def get_frame(self):
frame = self.game.getScreenGrayscale()
color_frame = self.game.getScreenRGB()
return frame, color_frame
def process_frame(self, frame):
#normalize
processed = np.reshape(frame, [np.prod(frame.shape), 1]) / 255.0
return processed
def game_finished(self):
return self.game.game_over()
def make_action(self, action_index):
reward = self.game.act(self.actions[action_index])
#update visitation map
self.coordinates = (self.game.game.getGameState()['player_x'],
10)
self.visitation_map[self.timer] = self.coordinates
self.timer += 1
return reward
def construct_visitation_map(self):
image = np.uint8(np.zeros((11,self.width, 4)))
image = Image.fromarray(image)
image = image.convert("RGBA")
pixels = image.load()
opacity = 100
increase = 20
for timestep in self.visitation_map:
coordinate = self.visitation_map[timestep]
if pixels[coordinate[0], coordinate[1]] == (0,0,0,0):
pixels[coordinate[0],coordinate[1]] = (255,0,0, int(opacity))
else:
value = tuple(sum(x) for x in zip(pixels[coordinate[0],
coordinate[1]],
(0, 0, 0, int(increase))))
pixels[coordinate[0], coordinate[1]] = value
#mark start and end positions
coordinate = self.visitation_map[0]
pixels[coordinate[0], coordinate[1]/2] = (0,255,0,255)
pixels[coordinate[0], coordinate[1]/2 - 1] = (0, 255, 0, 255)
coordinate = self.visitation_map[len(self.visitation_map) - 1]
# rewrite coordinate
pixels[coordinate[0], coordinate[1]/2] = (0, 0, 255, 255)
return image
#TODO GENERIC PLE WRAPPER
class RaycastMazeWrapper:
def __init__(self, width):
'''
@width : width of game window
'''
self.game = None
self.actions = None
# Maximum 1000 steps in maze
self.max_game_len = 500
self.frames_no = 0
# Create game env
raycast = RaycastMaze(width=width, height=width, map_size=6)
self.game = self.set_maze_game_setup(raycast)
def set_maze_game_setup(self, game):
'''
@game : game instance
'''
p = PLE(game, display_screen=False)
#In some games, doing nothing is a valid action
#in a maze, it is not
self.actions = p.getActionSet()[:-1]
p.init()
return p
def restart_game(self):
self.game.reset_game()
frame_skip = random.randint(0, 30)
# Randomize start
for i in range(frame_skip):
reward = self.make_action(random.choice(range(len(self.actions))))
def get_frame(self):
frame = self.game.getScreenGrayscale()
color_frame = self.game.getScreenRGB()
return frame, color_frame
def process_frame(self, frame):
'''
@frame : frame to be processed
'''
# normalize
processed = np.reshape(frame, [np.prod(frame.shape), 1]) / 255.0
return processed
def game_finished(self):
return self.game.game_over()
def make_action(self, action_index):
'''
@action_index : index of action
'''
reward = self.game.act(self.actions[action_index])
return reward
class PixelcopterWrapper:
def __init__(self, width):
'''
@width : width of game window
'''
self.game = None
self.actions = None
# Maximum 1000 steps in maze
self.max_game_len = 300
self.frames_no = 0
# Create game env
raycast = Pixelcopter_v2(width=width, height=width)
self.game = self.set_maze_game_setup(raycast)
def construct_visitation_map(self):
return None
def set_maze_game_setup(self, game):
'''
@game : game instance
'''
p = PLE(game, display_screen=True)
self.actions = p.getActionSet()
p.init()
return p
def restart_game(self):
self.game.reset_game()
#don't randomize start since it will most likely end the game
#frame_skip = random.randint(0, 30)
# Randomize start
#for i in range(frame_skip):
# reward = self.make_action(random.choice(range(len(self.actions))))
def get_frame(self):
frame = self.game.getScreenGrayscale()
color_frame = self.game.getScreenRGB()
return frame, color_frame
def process_frame(self, frame):
'''
@frame : frame to be processed
'''
# normalize
processed = np.reshape(frame, [np.prod(frame.shape), 1]) / 255.0
return processed
def game_finished(self):
return self.game.game_over()
def make_action(self, action_index):
'''
@action_index : index of action
'''
reward = self.game.act(self.actions[action_index])
return reward
class GameWrapper:
def __init__(self, game_name, window_width):
'''
@game_name : name of required game
@window_width : width of window/image to be used
'''
self.game = None
if game_name == 'Doom':
self.game = DoomWrapper(window_width)
self.game.name = 'Doom'
if game_name == 'Catcher':
self.game = CatcherWrapper(window_width)
self.game.name = 'Catcher'
if game_name == 'Maze':
self.game = RaycastMazeWrapper(window_width)
self.game.name = 'Maze'
if game_name == 'Copter':
self.game = PixelcopterWrapper(window_width)
self.game.name = 'Copter'
if game_name == 'LabMaze':
self.game = LabWrapper(window_width)
self.game.name = 'LabMaze'
def get_game(self):
return self.game