def output_winner(self):
SingleExperiment.output_winner(self)
self.exp_runner.render = True
self.exp_runner.run(self.winner, self.learning_config)
self.exp_runner.env.produce_trajectory(self.base_directory +
'trajectory' +
str(self.exp_name) + '.png')
self.exp_runner.render = False
def __init__(self,
learning_config,
num_generations,
controller_keeper,
exp_name='',
num_trails=1,
base_directory='',
cntrl_draw_func=None):
""" Note the additional parameter genome_encoder, which encodes a genome for sending. """
SingleExperiment.__init__(self, learning_config, num_generations,
exp_name, num_trails, base_directory,
cntrl_draw_func)
self.controllers = controller_keeper
self.genome_encoder = self.controllers.genome_encoder
# Initialise the variables needed to communicate with the others trough ROS.
self.gen_hash = 0
rospy.init_node('evolutionary_algorithm', anonymous=True)
from examples.experiment_functions import SMSwarmExperimentRunner
from examples.experiment_template import SingleExperiment
# Important variables.
from neat.state_machine_genome import StateMachineGenome
experiment_name = 'SM_4_reference_block_long'
num_steps = 3000
num_generations = 200
num_runs = 1
config_name = 'config-state_machine'
if __name__ == '__main__':
env = gym.make('tiling-pattern11x11-block-v0')
runner = SMSwarmExperimentRunner(env, num_steps)
# Create learning configuration.
local_dir = os.path.dirname(__file__)
config_path = os.path.join(local_dir, config_name)
config = neat.Config(StateMachineGenome, neat.DefaultReproduction,
neat.DefaultSpeciesSet, neat.DefaultStagnation,
config_path)
# Create and run experiment.
experiment = SingleExperiment(config, runner, num_generations,
experiment_name)
for i in range(num_runs):
experiment.run(experiment_name + str(i))