def generate(generations, population, nn_param_chices, dataset):
genetic = Genetic(nn_param_chices)
gans = genetic.create_population(population)
for i in range(generations):
train_gan(gans, dataset)
average_accuracy = get_average_accuracy(gans)
if i != generations - 1:
gans = genetic.evolve(gans)
gans = sorted(gans, key=lambda x: x.accuracy, reverse=True)
print("")
for r in robots_sorted[:5]:
print ("{:.2f}%\t{:.2f}%\t{:.2f}%\t{}".format(100*r.get_fitness(), 100*r.get_fitness_distances(), 100*r.get_fitness_sensors(), r.id))
print ("#"*40)
print ("\nEnded generation {} after {} steps".format(generation, step))
print ("Best robot got to distance {} and fitness {:.2f}%".format(closest_distance, 100*best_fitness))
generation_scores.append([closest_distance, best_fitness])
print ("#"*40)
generation += 1
# input("press enter to continue")
#get robots gains and fitness value to run genetic algorithm
results = list()
for r in robots:
gains = r.control_unit.get_gains()
fitness = r.get_fitness()
robot_result = {"gains":gains, "fitness":fitness, "id":r.id}
results.append(robot_result)
g = Genetic(results)
new_gains = g.evolve(keep_percent=0.20)
for i, n in enumerate(neurals):
n.set_gains(new_gains[i]["gains"])
