-
Notifications
You must be signed in to change notification settings - Fork 3
/
behavior.py
424 lines (360 loc) · 15.1 KB
/
behavior.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
import random
import numpy
"""""
*first byte*
sensory inputs
01 - blocked forward 02 - blocked back 03 - blocked left 04 - blocked right 05 - population density
06 - last movement was x 07 - last movement was y 08 - border distance north 09 - border distance east
0A - border distance south 0B - border distance west 0C - nearest border distance 0D - current loc. north
0E - current loc. east 0F - current loc. south 10 - current loc. west
internal neurons
11 - n_1 12 - n_2 13 - n_3
first byte of genome ranges from 01 to 13 (mod 19 + 1)
reserve 00 for detects food, which will have special connections to move towards the food.
*second byte*
internal neurons
01 - n_1 02 - n_2 03 - n_3
action outputs
04 - move forward 05 - move random 06 - move backwards 07 - move left
08 - move right 09 - move north 0A - move east 0b - move south 0C - move west
second byte ranges from 01 to 0C (mod 12 + 1)
reserve 00 for move to food, which is reserved for special sensory input
*third byte*
this will determine the strength of connection, ranging from -4.0 to 4.0
00 will represent the extreme of -4.000
AA will represent the extreme of 4.000
*fourth byte*
currently unused, will keep open for potential down the road since there is no 24-bit integers
"""""
# worker functions
# byte_to_float is used for byte 3 to convert to a float
# int_to_hex_string is used for random genome generation and ensures that the output is of desired length
# extract_hex_sub is used to get the bytes from the hex string
def byte_to_float(byte):
# Convert the input byte (hex string) to an integer
value = int(byte, 16)
# Calculate the float value within the range -4.0 to 4.0
min_range = -4.0
max_range = 4.0
float_value = min_range + (max_range - min_range) * (value / 255.0)
return float_value
def int_to_hex(number):
# Convert the integer to a hexadecimal string with '0x' prefix, and remove the prefix
hex_string = hex(number)[2:]
# Ensure the string is exactly two characters long by padding with '0' if needed
hex_string = hex_string.zfill(2)
return hex_string
def hex_to_int(hex_str):
try:
return int(hex_str, 16)
except ValueError:
return None
def extract_hex_sub(input_string, start_index, end_index):
if start_index < 0 or end_index > len(input_string):
return "Invalid indices"
substring = input_string[start_index:end_index]
return substring
def generate_genome():
# change numbers as program expands
num_inputs = 16
num_outputs = 9
num_int_neurons = 3
genome = ""
# first byte
genome += int_to_hex(random.randint(1, num_inputs + num_int_neurons)) # random number, then convert to 2 char hex string
# second byte
genome += int_to_hex(random.randint(1, num_outputs + num_int_neurons)) # same as before
# third byte
# as of right now, we are going to weight connections positively, since we would like to see action
# may need to revert later
genome += int_to_hex(random.randint(55, 255))
# fourth byte
genome += int_to_hex(random.randint(0, 255))
return genome
def sensor_switch_handler(properties, sensor):
match sensor:
# blocked forwards
case 1:
return properties[3] * .5
# blocked backwards
case 2:
return properties[4] * .5
# blocked left
case 3:
return properties[5] * .5
# blocked right
case 4:
return properties[6] * .5
# population density
case 5:
return properties[7] * .16
# last movement was in x dir
case 6:
if properties[2] == 1 or properties[2] == 3:
return 1.0
return 0.0
# last movement was in y dir
case 7:
if properties[2] == 0 or properties[2] == 2:
return 1.0
return 0.0
# border distance north
# formula is (dist from north border / grid height)^2
case 8:
grid_height = 50
dist_north = properties[1]
return (dist_north / grid_height) * (dist_north / grid_height)
# border distance east
case 9:
grid_height = 50
dist_east = grid_height - properties[0]
return (dist_east / grid_height) * (dist_east / grid_height)
# border distance south
case 10:
grid_height = 50
dist_south = grid_height - properties[1]
return (dist_south / grid_height) * (dist_south / grid_height)
# border distance west
case 11:
grid_height = 50
dist_west = properties[0]
return (dist_west / grid_height) * (dist_west / grid_height)
# nearest border distance
case 12:
grid_height = 50
nearest_border = max(properties[0], properties[1], grid_height - properties[1],
grid_height - properties[0])
return (nearest_border / grid_height) * .8 # arbitrary value to prevent overpowering
# current location north
# formula is (dist from south border / grid height)^2
case 13:
grid_height = 50
dist_south = grid_height - properties[1]
return (dist_south / grid_height) * (dist_south / grid_height)
# current location east
case 14:
grid_height = 50
dist_west = properties[0]
return (dist_west / grid_height) * (dist_west / grid_height)
# current location south
case 15:
grid_height = 50
dist_north = properties[1]
return (dist_north / grid_height) * (dist_north / grid_height)
# current location west
case 16:
grid_height = 50
dist_west = grid_height - properties[0]
return (dist_west / grid_height) * (dist_west / grid_height)
case _:
print("sensor_switch_handler defaulting")
# if_neuron_triggers takes in a creature's properties and returns which neurons triggers
def if_neuron_triggers(properties, input_sensors):
num_inputs = 16
# make array to be returned
ret = [0.0] * (num_inputs + 1)
for sensor in input_sensors:
ret[sensor] = sensor_switch_handler(properties, sensor)
return ret
# create_active_map takes in a creature's genome and active sensors to create an adjacency map of network
# the key is the input sensor and the value is a list of pairs, with format (output index, connection strength)
def create_active_map(genome, active_sensors):
num_inputs = 16
num_outputs = 12
num_internal_neurons = 3
# create dict to be returned
ret = {}
# populate adjacency map with network
for k in genome:
# define genome connection in adjacency map
input_node = hex_to_int(extract_hex_sub(k, 0, 2))
conn_str = byte_to_float(extract_hex_sub(k, 4, 6))
if input_node not in ret:
ret[input_node] = []
# to_add is pair with format (output index, connection strength)
to_add = [hex_to_int(extract_hex_sub(k, 2, 4)), conn_str]
ret[input_node].append(to_add)
# filter internal connections with no output
# TODO: create function to filter internal connections with no meaningful output
return ret
# creature_behavior_output takes in a creature and returns an array of output probabilities
def creature_behavior_output(properties, genome):
# change numbers as program expands
num_inputs = 16
num_outputs = 9
num_internal_neurons = 3
# create temp variables
active_sensors = []
output_sensors = [0.0] * (num_outputs + 1)
internal_neurons = [0.0] * (num_internal_neurons + 1)
# this loop lists all sensory signals
for gene in genome:
# get input sensor from genome
temp = hex_to_int(extract_hex_sub(gene, 0, 2))
if temp <= num_inputs: # if temp is sensory
# sensory neuron can trigger a connection, add to list
active_sensors += [temp]
# print(active_sensors)
# handle all active sensors
input_sensors = if_neuron_triggers(properties, active_sensors)
# print(input_sensors)
# active_map is a key-value data structure where the key is the input sensor
# the value is the list of adjacent output nodes (can include input sensor)
active_map = create_active_map(genome, input_sensors)
active_map = sorted(active_map.items())
print(active_map)
# sort so internal neurons are processed last
# use active_map and input_sensors to calculate sensory input in neural network
for input_index, output_list in active_map:
for key, conn_str in output_list: # for each output attached to the input
# split based on internal or output
if key <= num_internal_neurons: # output is internal
if input_index < num_inputs:
internal_neurons[key - 1] += (input_sensors[input_index] * conn_str) # add signal to internal
else:
internal_neurons[key - 1] += (internal_neurons[input_index - num_inputs - 1] * conn_str)
else: # output is action
if input_index < num_inputs:
output_sensors[key - num_internal_neurons] += (input_sensors[input_index] * conn_str)
else:
output_sensors[key - num_internal_neurons] += (internal_neurons[input_index - num_inputs] * conn_str)
# calculate outputs
return numpy.tanh(output_sensors)
# movement_output takes in the list of output_sensors and properties of the organism
# the output is an array of size 4 with each coordinating for N E S W
def movement_output(output_sensors, properties):
output_index = numpy.array(output_sensors).argmax()
output_array = [0, 0, 0, 0]
match output_index:
# move forward
case 1:
output_array[properties[2]] = 1
return output_array
# move random
case 2:
random_index = random.randint(0, 3)
output_array[random_index] = 1
return output_array
# move backwards
case 3:
output_array[(properties[2] + 2) % 4] = 1
return output_array
# move left
case 4:
output_array[(properties[2] + 3) % 4] = 1
return output_array
# move right
case 5:
output_array[(properties[2] + 1) % 4] = 1
return output_array
# move north
case 6:
output_array[0] = 1
return output_array
# move east
case 7:
output_array[1] = 1
return output_array
# move south
case 8:
output_array[2] = 1
return output_array
# move west
case 9:
output_array[3] = 1
return output_array
# default case (move forward)
case _:
output_array[properties[2]] = 1
return output_array
# clean_genome takes in a genome string and ensures all values are within allotted ranges
def clean_genome(genome):
# change numbers as program expands
num_inputs = 16
num_outputs = 9
num_internal_neurons = 3
ret = []
for gene in genome:
new_gene = ""
# first byte (input)
to_add = extract_hex_sub(gene, 0, 2)
to_add = hex_to_int(to_add) % (num_inputs + num_internal_neurons)
# this is a temporary fix, since 00 is reserved for food detected
if to_add == 0:
to_add = 1
new_gene += int_to_hex(to_add)
# second byte
to_add = extract_hex_sub(gene, 2, 4)
to_add = hex_to_int(to_add) % (num_outputs + num_internal_neurons)
# this is a temporary fix, since 00 is reserved for move to food
if to_add == 0:
to_add = 1
new_gene += int_to_hex(to_add)
# third and fourth byte (these values are converted and do not need to be modulo'd)
new_gene += extract_hex_sub(gene, 4, 8)
ret.append(new_gene)
return ret
# mutate_genome takes in a genome string and a float mutation probability
# if the probability succeeds, that particular string in the genome will mutate a random bit
def mutate_genome(genome, mutation_chance):
for g in range(len(genome)):
# gene mutates
if random.randint(0, 100) / 100.0 <= mutation_chance:
mutate_index = random.randint(0, len(genome[g]) - 1)
mutate_char = int_to_hex(random.randint(0, 15))
temp = list(genome[g])
return genome
# reproduce_genome takes in two genome strings and an integer to declare which mode we are using
# 0 is full string genome, 1 is segmented genome, 2 is random genome
# the function returns a new genome that is a descendant of the two given as parameters
def reproduce_genome(parent1, parent2, mode):
assert len(parent1) == len(parent2)
new_gen = []
if mode == 0: # full string genome
for gene in range(len(parent1)):
# parent 1
if gene % 2 == 0:
new_gen.append(parent1[gene])
# parent 2
else:
new_gen.append(parent2[gene])
return new_gen
elif mode == 1: # segmented string genome
parent_ind = 0; # parent_ind keeps track of which parent the segment is being taken from
parent_list = [parent1, parent2]
for gene in range(len(parent1)): # for each gene in the genome
to_add = ""
start_ind = 0;
end_ind = 2;
for seg in range(4): # for each segment in the gene
to_add += (extract_hex_sub(parent_list[parent_ind][gene], start_ind, end_ind))
# increment vars
parent_ind = (parent_ind + 1) % 2
start_ind += 2
end_ind += 2
new_gen.append(to_add)
return new_gen
elif mode == 2: # random string genome
for gene in range(len(parent1)):
to_add = ""
for char in range(len(parent1[0])):
parent_index = random.randint(1, 8) % 2
if parent_index == 0:
to_add += (extract_hex_sub(parent1[gene], char, char + 1))
elif parent_index == 1:
to_add += (extract_hex_sub(parent2[gene], char, char + 1))
new_gen.append(to_add)
return new_gen
creature_genome = []
# x coord, y coord, last moved direction (0 is north, 1 is east, 2 is south, 3 is west)
# blocked forwards, back, left, right (boolean), population count in sensory range (integer),
# TODO: when merge with main branch, use the grid size in simulation.toml instead of hard coded
creature_properties = [22, 24, 2, 0, 1, 0, 1, 1]
for i in range(4):
creature_genome.append(generate_genome())
output_names = ['move to food', 'move forward', 'move random', 'move backwards', 'move left', 'move right',
'move north', 'move east', 'move south', 'move west']
output_vec = creature_behavior_output(creature_properties, creature_genome)
for f in range(len(output_vec)):
print(output_names[f], ": ", output_vec[f])
print(movement_output(output_vec, creature_properties))