def mutate(self, individual):
genome = self.convert_to_genome(individual.brain.weights)
weight_to_mutate = random.randint(0, len(genome) - 1)
# genome[weight_to_mutate] = genome[weight_to_mutate] * random.uniform(0.5, 1.5)
genome[weight_to_mutate] = np.random.randn()
new_weights = self.convert_to_weight(genome, individual.brain.weights)
return Robot(50, 300, 8, 360, 9, all, new_weights, own_weights=True)
def create_child(self, parent1, parent2): # Uniform crossover
parent1_genome = self.convert_to_genome(
parent1.brain.weights
) # but is DNA actually being updated over time (over the generations)
parent2_genome = self.convert_to_genome(parent2.brain.weights)
child_genome = []
for i in range(len(parent1_genome)):
if random.random() > 0.5:
child_genome.append(parent1_genome[i])
else:
child_genome.append(parent2_genome[i])
child_weights = self.convert_to_weight(child_genome,
parent1.brain.weights)
return Robot(50, 300, 8, 360, 9, all, child_weights, own_weights=True)
from Sensor import Robot
from Pedestrian import all, Darwin
import numpy as np
a = Robot(200, 300, 8, 360, 9, all)
b = Robot(200, 300, 8, 360, 9, all)
print(a, b)
print(a.fitness)
robot_array = [a, b]
print([i.fitness for i in robot_array])
darwin = Darwin(robot_array, 2, 1)
genome = darwin.convert_to_genome(a.DNA)
weights = darwin.convert_to_weight(genome, a.DNA)
print(np.shape(weights), np.shape(a.DNA))
c = darwin.create_child(a, b)
# print(np.array_equal(a.DNA, a.brain.weights))
# print(a.DNA)
# print(a.brain.weights)
a.DNA = c
for i in range(3):
print(np.array_equal(a.DNA[i], a.brain.weights[i]))
starting_x = My_Trajectory_Dict[pedestrian][0][0]
starting_y = My_Trajectory_Dict[pedestrian][0][1]
all_pedestrians.append(
Pedestrian(starting_x, starting_y, 8, pedestrian,
My_Trajectory_Dict[pedestrian]))
all = Manager(all_pedestrians, 15)
# obstacleArray = [Obstacle('Circle', (500, 300), (0, 0, 255), 0, 100), Obstacle('Circle', (200, 300), (0, 255, 0), 0, 75)]
population_size = 50
elitism = 4
robots = []
for i in range(population_size):
robots.append(Robot(100, 300, 8, 360, 9, all, set_weights=None))
# some class here maybe to manage all robots
class Darwin:
def __init__(self, robot_array, elitism, mutation_rate):
self.robot_array = robot_array
self.generation = 0
self.population_size = population_size
self.elitism = elitism
self.mutation_rate = mutation_rate
self.best_fitness = 0
self.number_of_parents = 4
self.dead_count = 0
self.x_data = []
self.y_data = []