def build_env(enable_randomizer,
enable_rendering,
version=0,
mode='train',
control_mode='torque'):
sim_params = locomotion_gym_config.SimulationParameters()
sim_params.enable_rendering = enable_rendering
gym_config = locomotion_gym_config.LocomotionGymConfig(
simulation_parameters=sim_params)
robot_class = laikago.Laikago
sensors = [
HistoricSensorWrapper(NormalizeSensorWrapper(
wrapped_sensor=robot_sensors.MotorAngleSensor(
num_motors=laikago.NUM_MOTORS)),
num_history=3),
HistoricSensorWrapper(NormalizeSensorWrapper(
wrapped_sensor=robot_sensors.MotorVelocitiySensor(
num_motors=laikago.NUM_MOTORS)),
num_history=3),
HistoricSensorWrapper(
NormalizeSensorWrapper(wrapped_sensor=robot_sensors.IMUSensor()),
num_history=3),
HistoricSensorWrapper(NormalizeSensorWrapper(
wrapped_sensor=robot_sensors.ToeTouchSensor(laikago.NUM_LEGS)),
num_history=3),
HistoricSensorWrapper(NormalizeSensorWrapper(
wrapped_sensor=environment_sensors.LastActionSensor(
num_actions=laikago.NUM_MOTORS)),
num_history=3)
]
task = eval('task.StandupTaskV{}'.format(version))(mode=mode)
randomizers = []
if enable_randomizer:
randomizer = controllable_env_randomizer_from_config.ControllableEnvRandomizerFromConfig(
verbose=False)
randomizers.append(randomizer)
if control_mode == 'torque':
motor_control_mode = MotorControlMode.TORQUE
else:
motor_control_mode = MotorControlMode.POSITION
init_pose = 'stand'
env = locomotion_gym_env.LocomotionGymEnv(
gym_config=gym_config,
robot_class=robot_class,
motor_control_mode=motor_control_mode,
init_pose=init_pose,
env_randomizers=randomizers,
robot_sensors=sensors,
task=task)
env = observation_dictionary_to_array_wrapper.ObservationDictionaryToArrayWrapper(
env)
return env
def build_env(enable_randomizer,
enable_rendering,
mode='train',
version=0,
force=True):
sim_params = locomotion_gym_config.SimulationParameters()
sim_params.enable_rendering = enable_rendering
gym_config = locomotion_gym_config.LocomotionGymConfig(
simulation_parameters=sim_params)
robot_class = laikago.Laikago
sensors = [
sensor_wrappers.HistoricSensorWrapper(
wrapped_sensor=robot_sensors.MotorAngleSensor(
num_motors=laikago.NUM_MOTORS),
num_history=3),
sensor_wrappers.HistoricSensorWrapper(
wrapped_sensor=robot_sensors.MotorVelocitiySensor(
num_motors=laikago.NUM_MOTORS),
num_history=3),
sensor_wrappers.HistoricSensorWrapper(
wrapped_sensor=robot_sensors.IMUSensor(), num_history=3),
sensor_wrappers.HistoricSensorWrapper(
wrapped_sensor=environment_sensors.LastActionSensor(
num_actions=laikago.NUM_MOTORS),
num_history=3),
height_sensor.HeightSensor(random_height=True)
]
task = eval('standupheight_task.StandupheightTaskV{}'.format(version))(
mode=mode, force=force)
randomizers = []
if enable_randomizer:
randomizer = controllable_env_randomizer_from_config.ControllableEnvRandomizerFromConfig(
verbose=False)
randomizers.append(randomizer)
env = locomotion_gym_env.LocomotionGymEnv(gym_config=gym_config,
robot_class=robot_class,
env_randomizers=randomizers,
robot_sensors=sensors,
task=task)
env = observation_dictionary_to_array_wrapper.ObservationDictionaryToArrayWrapper(
env)
return env