def evaluate(self, genome, generation, genome_id):
"""Generate a game, test the genome and return its fitness.
Keyword arguments:
genome -- genome to test
generation -- number of the current generation
genome_id -- genome id in the current generation
"""
# create neural network from genome.
net = NEAT.NeuralNetwork()
genome.BuildPhenotype(net)
# create player and .
player = Dino_player_neat(net)
text = "generation : " + str(generation) + " || genome : " + str(
genome_id)
the_game = Dino(player, text)
fitness = the_game.on_execute(start_immediately=True)
return fitness
if len(arguments) != 0 and len(sys.argv) == 1:
return arguments
else:
return None
if __name__ == "__main__":
arguments = verify_arguments()
if arguments is None:
print(f'Usage: python {sys.argv[0]} [ --type (human|random|neat \
[--population <size>] [--generations <size>]) ] ')
elif "type" in arguments:
if arguments["type"] == "human":
player = Dino_player()
theDino = Dino(player)
theDino.on_execute()
elif arguments["type"] == "random":
player = Dino_player_random()
theDino = Dino(player)
theDino.on_execute()
else:
data = dict()
if "population" in arguments:
data["population_size"] = arguments["population"]
if "generations" in arguments:
data["generations"] = arguments["generations"]
cycle = NEAT_trainer(**data)
cycle.start_cycle()
else:
player = Dino_player()
theDino = Dino(player)