-
Notifications
You must be signed in to change notification settings - Fork 0
/
api.py
603 lines (523 loc) · 18.4 KB
/
api.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
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
# api.py
# parsons/15-nov-2017
#
# Version 6
#
# With acknowledgements to Jiaming Ke, who was the first to report the
# bug in corners and to spot the bug in the motion model.
#
# An API for use with the PacMan AI projects from:
#
# http://ai.berkeley.edu/
#
# This provides a simple way of controlling the way that Pacman moves
# and senses its world, to permit exercises with limited sensing
# ability and nondeterminism in sensing and action.
#
# As required by the licensing agreement for the PacMan AI we have:
#
# Licensing Information: You are free to use or extend these projects for
# educational purposes provided that (1) you do not distribute or publish
# solutions, (2) you retain this notice, and (3) you provide clear
# attribution to UC Berkeley, including a link to http://ai.berkeley.edu.
#
# Attribution Information: The Pacman AI projects were developed at UC Berkeley.
# The core projects and autograders were primarily created by John DeNero
# ([email protected]) and Dan Klein ([email protected]).
# Student side autograding was added by Brad Miller, Nick Hay, and
# Pieter Abbeel ([email protected]).
# The code here was written by Simon Parsons, based on examples from
# the PacMan AI projects.
from random import random
from pacman import Directions
import util
#
# Parameters
#
# Control visibility.
#
# If partialVisibility is True, Pacman will only see part of the
# environment.
partialVisibility = False
# The limits of visibility when visibility is partial
sideLimit = 1
hearingLimit = 2
visibilityLimit = 5
# Control determinism
#
# If nonDeterministic is True, Pacman's action model will be
# nonDeterministic.
nonDeterministic = False
# Probability that Pacman carries out the intended action:
directionProb = 1
#
# Sensing
#
def whereAmI(state):
# Returns an (x, y) pair of Pacman's position.
#
# This version says exactly where Pacman is.
# In later version this may be obfusticated.
return state.getPacmanPosition()
def legalActions(state):
# Returns the legal set of actions
#
# Just pulls this data out of the state. Function included so that
# all interactions are through this API.
return state.getLegalPacmanActions()
def ghosts(state):
# Returns a list of (x, y) pairs of ghost positions.
#
# This version returns the ghosts in positions that are visible
# and audible
return union(visible(state.getGhostPositions(),state), audible(state.getGhostPositions(),state))
def ghostStates(state):
# Returns the position of the ghsosts, plus an indication of
# whether or not they are scared/edible.
#
# The information is returned as a list of elements of the form:
#
# ((x, y), state)
#
# where "state" is 1 if the relevant ghost is scared/edible, and 0
# otherwise.
ghostStateInfo = state.getGhostStates()
ghostStates = []
for s in ghostStateInfo:
if s.scaredTimer > 0:
ghostStates.append((s.getPosition(), 1))
else:
ghostStates.append((s.getPosition(), 0))
return ghostStates
def ghostStatesWithTimes(state):
# Just as ghostStates(), but when the ghost is in scared/edible
# mode, "state" is a time value (how much longer the ghost will
# remain scared/edible) rather than 1.
ghostStateInfo = state.getGhostStates()
ghostStates = []
for s in ghostStateInfo:
ghostStates.append((s.getPosition(), s.scaredTimer))
return ghostStates
def capsules(state):
# Returns a list of (x, y) pairs of capsule positions.
#
# This version returns the capsule positions if they are within
# the distance limit.
#
# Capsules are visible if:
#
# 1) Pacman is moving and the capsule is in front of Pacman and
# within the visibilityLimit, or to the side of Pacman and within
# the sideLimit.
#
# 2) Pacman is not moving, and the capsule is within the visibilityLimit.
#
# In both cases, walls block the view.
return visible(state.getCapsules(), state)
def food(state):
# Returns a list of (x, y) pairs of food positions
#
# This version returns all the current food locations that are
# visible.
#
# Food is visible if:
#
# 1) Pacman is moving and the food is in front of Pacman and
# within the visibilityLimit, or to the side of Pacman and within
# the sideLimit.
#
# 2) Pacman is not moving, and the food is within the visibilityLimit.
#
# In both cases, walls block the view.
foodList= []
foodGrid = state.getFood()
width = foodGrid.width
height = foodGrid.height
for i in range(width):
for j in range(height):
if foodGrid[i][j] == True:
foodList.append((i, j))
# Return list of food that is visible
return visible(foodList, state)
def walls(state):
# Returns a list of (x, y) pairs of wall positions
#
# This version just returns all the current wall locations
# extracted from the state data. In later versions, this will be
# restricted by distance, and include some uncertainty.
wallList= []
wallGrid = state.getWalls()
width = wallGrid.width
height = wallGrid.height
for i in range(width):
for j in range(height):
if wallGrid[i][j] == True:
wallList.append((i, j))
return wallList
def corners(state):
# Returns the coordinates of the four corners of the state space.
#
# For harder exploration we could obfusticate this information.
corners=[]
wallGrid = state.getWalls()
width = wallGrid.width
height = wallGrid.height
corners.append((0, 0))
corners.append((width-1, 0))
corners.append((0, height-1))
corners.append((width-1, height-1))
return corners
#
# Acting
#
def makeMove(direction, legal):
# This version implements non-deterministic movement.
#
# Paacman has a probability of directionProb of moving in the
# specified direction, and 0.5*(1 - directionProb) of moving
# perpendicular to the specified direction. Any attempt to move in
# an direction that is not legal means Pacman stays in the same
# place.
#
# With the default setting of directionProb = 0.8, this is exactly
# the motion model we studied in the MDP lecture.
# If Pacman hasn't yet moved, then non-determinism plays no role in
# deciding what Pacman does:
if direction == Directions.STOP:
return direction
if nonDeterministic:
# Sample in the usual way to make Pacman move in the specified
# direction with probability directionProb.
#
# Otherwise make a different move.
sample = random()
if sample <= directionProb:
# Here the non-deterministic action selection says to
# return the original move, but we need to check it is
# legal in case we were passed an illegal action (because,
# for example, that was the MEU action).
#
# If the specified action is not legal, in this case we
# will not move.
if direction in legal:
return direction
else:
return Directions.STOP
else:
return selectNewMove(direction, legal)
else:
# When actions are deterministic, Pacman moves in the
# specified direction (with another check on legality).
if direction in legal:
return direction
else:
return Directions.STOP
#
# Details that you don't need to look at if you don't want to.
#
def distanceLimited(objects, state, limit):
# When passed a list of object locations, tests how far they are
# from Pacman, and only returns the ones that are within "limit".
pacman = state.getPacmanPosition()
nearObjects = []
for i in range(len(objects)):
if util.manhattanDistance(pacman,objects[i]) <= limit:
nearObjects.append(objects[i])
return nearObjects
def inFront(object, facing, state):
# Returns true if the object is along the corridor in the
# direction of the parameter "facing" before a wall gets in the
# way.
pacman = state.getPacmanPosition()
pacman_x = pacman[0]
pacman_y = pacman[1]
wallList = walls(state)
# If Pacman is facing North
if facing == Directions.NORTH:
# Check if the object is anywhere due North of Pacman before a
# wall intervenes.
next = (pacman_x, pacman_y + 1)
while not next in wallList:
if next == object:
return True
else:
next = (pacman_x, next[1] + 1)
return False
# If Pacman is facing South
if facing == Directions.SOUTH:
# Check if the object is anywhere due North of Pacman before a
# wall intervenes.
next = (pacman_x, pacman_y - 1)
while not next in wallList:
if next == object:
return True
else:
next = (pacman_x, next[1] - 1)
return False
# If Pacman is facing East
if facing == Directions.EAST:
# Check if the object is anywhere due East of Pacman before a
# wall intervenes.
next = (pacman_x + 1, pacman_y)
while not next in wallList:
if next == object:
return True
else:
next = (next[0] + 1, pacman_y)
return False
# If Pacman is facing West
if facing == Directions.WEST:
# Check if the object is anywhere due West of Pacman before a
# wall intervenes.
next = (pacman_x - 1, pacman_y)
while not next in wallList:
if next == object:
return True
else:
next = (next[0] - 1, pacman_y)
return False
def atSide(object, facing, state):
# Returns true if the object is in a side corridor perpendicular
# to the direction that Pacman is travelling.
pacman = state.getPacmanPosition()
# If Pacman is facing North or Sout, then objects to the side are to the
# East and West.
#
# These are objects that Pacman would see if it were facing East
# or West.
if facing == Directions.NORTH or facing == Directions.SOUTH:
# Check if the object is anywhere due North of Pacman before a
# wall intervenes.
if inFront(object, Directions.WEST, state) or inFront(object, Directions.EAST, state):
return True
else:
return False
# Similarly for other directions
if facing == Directions.WEST or facing == Directions.EAST:
# Check if the object is anywhere due North of Pacman before a
# wall intervenes.
if inFront(object, Directions.NORTH, state) or inFront(object, Directions.SOUTH, state):
return True
else:
return False
else:
return False
def visible(objects, state):
# When passed a list of objects, returns those that are visible to
# Pacman.
# This code creates partial observability by only returning some
# of the members of objects.
facing = state.getPacmanState().configuration.direction
visibleObjects = []
sideObjects = []
if facing != Directions.STOP:
# If Pacman is moving, visible objects are those in front of,
# and to the side (if there are any side corridors).
# Objects in front. Visible up to "visibilityLimit"
for i in range(len(objects)):
if inFront(objects[i], facing, state):
visibleObjects.append(objects[i])
visibleObjects = distanceLimited(visibleObjects, state, visibilityLimit)
# Objects to the side. Visible up to "sideLimit"
for i in range(len(objects)):
if atSide(objects[i], facing, state):
sideObjects.append(objects[i])
sideObjects = distanceLimited(sideObjects, state, sideLimit)
# Combine lists.
visibleObjects = visibleObjects + sideObjects
else:
# If Pacman is not moving, they can see in all directions.
#
# Unfortunately facing will never have value Directions.STOP
# after the first move is made, so this code will not run
# after the first move :-(
for i in range(len(objects)):
if inFront(objects[i], Directions.NORTH, state):
visibleObjects.append(objects[i])
if inFront(objects[i], Directions.SOUTH, state):
visibleObjects.append(objects[i])
if inFront(objects[i], Directions.EAST, state):
visibleObjects.append(objects[i])
if inFront(objects[i], Directions.WEST, state):
visibleObjects.append(objects[i])
visibleObjects = distanceLimited(visibleObjects, state, visibilityLimit)
# If we return visibleObjects, we have partial observability. If
# we return objects, then we have full observability.
if partialVisibility:
return visibleObjects
else:
return objects
def audible(ghosts, state):
# A ghost is audible if it is any direction and less than
# "hearingLimit" away.
return distanceLimited(ghosts, state, hearingLimit)
def union(a, b):
# return the union of two lists
#
# From https://www.saltycrane.com/blog/2008/01/how-to-find-intersection-and-union-of/
#
return list(set(a) | set(b))
def selectNewMove(direction, legal):
# This function is called if Pacman isn't moving in the specified
# direction. Need to pick another legal action.
# Pick with 50% probability between the two perpendicular
# possibilities.
sample = random()
if sample <= 0.5:
left = True
else:
left = False
# If chosen direction is North, then pick between West (left) and
# East. If these moves are legal, make them, otherwise don't move.
if direction == Directions.NORTH:
if left:
if Directions.WEST in legal:
return Directions.WEST
else:
return Directions.STOP
else:
if Directions.EAST in legal:
return Directions.EAST
else:
return Directions.STOP
# If chosen direction is EAST
if direction == Directions.EAST:
if left:
if Directions.NORTH in legal:
return Directions.NORTH
else:
return Directions.STOP
else:
if Directions.SOUTH in legal:
return Directions.SOUTH
else:
return Directions.STOP
# If chosen direction is SOUTH
if direction == Directions.SOUTH:
if left:
if Directions.EAST in legal:
return Directions.EAST
else:
return Directions.STOP
else:
if Directions.WEST in legal:
return Directions.WEST
else:
return Directions.STOP
# If chosen direction is WEST
if direction == Directions.WEST:
if left:
if Directions.SOUTH in legal:
return Directions.SOUTH
else:
return Directions.STOP
else:
if Directions.NORTH in legal:
return Directions.NORTH
else:
return Directions.STOP
print "Why am I here?"
#
# Bits for the machine learning task
def getFeatureVector(state):
# Returns local information about the environment in the form of a
# feature vector
features = []
xLoc = state.getPacmanPosition()[0]
yLoc = state.getPacmanPosition()[1]
#Are there walls around Pacman?
wallGrid = state.getWalls()
if wallGrid[xLoc][yLoc+1] == True:
features.append(1)
else:
features.append(0)
if wallGrid[xLoc+1][yLoc] == True:
features.append(1)
else:
features.append(0)
if wallGrid[xLoc][yLoc-1] == True:
features.append(1)
else:
features.append(0)
if wallGrid[xLoc-1][yLoc] == True:
features.append(1)
else:
features.append(0)
# Is there food around Pacman?
foodGrid = state.getFood()
if foodGrid[xLoc][yLoc+1] == True:
features.append(1)
else:
features.append(0)
if foodGrid[xLoc+1][yLoc] == True:
features.append(1)
else:
features.append(0)
if foodGrid[xLoc][yLoc-1] == True:
features.append(1)
else:
features.append(0)
if foodGrid[xLoc-1][yLoc] == True:
features.append(1)
else:
features.append(0)
# Are there ghosts in any of the eight squares around Pacman
ghosts = state.getGhostPositions()
facing = state.getPacmanState().configuration.direction
visibleGhost = False
for i in range(len(ghosts)):
if ghosts[i] == (xLoc-1, yLoc+1):
features.append(1)
else:
features.append(0)
if ghosts[i] == (xLoc, yLoc+1):
features.append(1)
else:
features.append(0)
if ghosts[i] == (xLoc+1, yLoc+1):
features.append(1)
else:
features.append(0)
if ghosts[i] == (xLoc-1, yLoc):
features.append(1)
else:
features.append(0)
if ghosts[i] == (xLoc+1, yLoc):
features.append(1)
else:
features.append(0)
if ghosts[i] == (xLoc-1, yLoc-1):
features.append(1)
else:
features.append(0)
if ghosts[i] == (xLoc, yLoc-1):
features.append(1)
else:
features.append(0)
if ghosts[i] == (xLoc+1, yLoc-1):
features.append(1)
else:
features.append(0)
# Is there a ghost in front of Pacman?
for i in range(len(ghosts)):
if inFront(ghosts[i], facing, state):
visibleGhost = True
if visibleGhost:
features.append(1)
else:
features.append(0)
return features
def getFeaturesAsString(state):
# Returns local information about the environment in the form of a
# string
features = ""
featureVector = getFeatureVector(state)
for i in range(len(featureVector)):
if featureVector[i] == 1:
features += "1"
else:
features += "0"
return features
#def featuresAsString(featureString):
# features = []
# for i in range(len(featureString)