def argsparser():
"""
Returns argument parser for furniture assembly environment.
"""
parser = argparse.ArgumentParser("Demo for IKEA Furniture Assembly Environment")
parser.add_argument('--seed', type=int, default=123)
parser.add_argument('--debug', type=str2bool, default=False)
import config.furniture as furniture_config
furniture_config.add_argument(parser)
args = parser.parse_args()
return args
def argsparser():
parser = argparse.ArgumentParser("Demo for furniture assembly with Baxter")
parser.add_argument('--seed', type=int, default=123)
parser.add_argument('--debug', type=str2bool, default=False)
parser.add_argument('--record_video', type=str2bool, default=False)
parser.add_argument('--video_name',
type=str,
default='FurnitureBaxterAssemblyEnv_test.mp4')
parser.add_argument('--verbose', type=str2bool, default=True)
parser.add_argument('--visualize_samples', type=str2bool, default=False)
parser.add_argument('--mirror', type=str2bool, default=False)
import config.furniture as furniture_config
furniture_config.add_argument(parser)
args = parser.parse_args()
return args
def main():
import argparse
import config.furniture as furniture_config
from util import str2bool
parser = argparse.ArgumentParser()
furniture_config.add_argument(parser)
# change default config for Baxter
parser.add_argument('--seed', type=int, default=123)
parser.add_argument('--debug', type=str2bool, default=False)
parser.set_defaults(render=True)
config, unparsed = parser.parse_known_args()
# create an environment and run manual control of Baxter environment
env = FurnitureBaxterAssemblyEnv(config)
env.plan_controller_compositions(config)
def main():
import argparse
import config.furniture as furniture_config
from util import str2bool
parser = argparse.ArgumentParser()
furniture_config.add_argument(parser)
# change default config for Cursors
parser.add_argument('--seed', type=int, default=123)
parser.add_argument('--debug', type=str2bool, default=False)
parser.set_defaults(render=True)
config, unparsed = parser.parse_known_args()
# create an environment and run manual control of Cursor environment
env = FurnitureCursorEnv(config)
env.run_manual(config)
def argsparser():
"""
Returns argument parser for furniture assembly environment.
"""
import argparse
from util import str2bool
parser = argparse.ArgumentParser("Demo for IKEA Furniture Assembly Environment")
parser.add_argument('--env_id', type=str, default='furniture-baxter-v0')
parser.add_argument('--num_env', type=int, default=1)
parser.add_argument('--seed', type=int, default=123)
parser.add_argument('--debug', type=str2bool, default=False)
import config.furniture as furniture_config
furniture_config.add_argument(parser)
parser.set_defaults(visual_ob=True)
args = parser.parse_args()
return args
def main():
import argparse
import config.furniture as furniture_config
from util import str2bool
parser = argparse.ArgumentParser()
furniture_config.add_argument(parser)
# change default config for Baxter
parser.add_argument('--seed', type=int, default=123)
parser.add_argument('--debug', type=str2bool, default=False)
parser.set_defaults(render=True)
parser.set_defaults(debug=True)
# fix the initialization of furniture parts
parser.set_defaults(fix_init=True)
config, unparsed = parser.parse_known_args()
# create an environment and run manual control of Baxter environment
env = FurnitureBaxterEnv(config)
env.run_manual(config)
def argparser():
parser = argparse.ArgumentParser(
'IKEA Furniture Assembly Environment',
formatter_class=argparse.ArgumentDefaultsHelpFormatter)
# environment
parser.add_argument('--env',
type=str,
default='FurnitureBaxterEnv',
choices=[
'FurnitureBaxterEnv', 'FurnitureSawyerEnv',
'FurnitureCursorEnv', 'FurnitureBaxterBlockEnv'
],
help='Environment name')
parser.add_argument('--env_args', type=str, default=None)
# furniture
import config.furniture as furniture_config
furniture_config.add_argument(parser)
# training algorithm
parser.add_argument('--algo',
type=str,
default='sac',
choices=['sac', 'ppo'])
parser.add_argument('--policy',
type=str,
default='mlp',
choices=['mlp', 'manual'])
# vanilla rl
parser.add_argument('--rl_hid_size', type=int, default=64)
parser.add_argument('--rl_activation',
type=str,
default='relu',
choices=['relu', 'elu', 'tanh'])
parser.add_argument('--tanh_policy', type=str2bool, default=True)
# observation normalization
parser.add_argument('--ob_norm', type=str2bool, default=True)
parser.add_argument('--max_ob_norm_step', type=int, default=int(1e7))
parser.add_argument('--clip_obs',
type=float,
default=200,
help='the clip range of observation')
parser.add_argument(
'--clip_range',
type=float,
default=5,
help='the clip range after normalization of observation')
# off-policy rl
parser.add_argument('--buffer_size',
type=int,
default=int(1e3),
help='the size of the buffer')
parser.add_argument('--discount_factor',
type=float,
default=0.99,
help='the discount factor')
parser.add_argument('--lr_actor',
type=float,
default=3e-4,
help='the learning rate of the actor')
parser.add_argument('--lr_critic',
type=float,
default=3e-4,
help='the learning rate of the critic')
parser.add_argument('--polyak',
type=float,
default=0.995,
help='the average coefficient')
# training
parser.add_argument('--is_train', type=str2bool, default=True)
parser.add_argument('--num_batches',
type=int,
default=50,
help='the times to update the network per epoch')
parser.add_argument('--batch_size',
type=int,
default=256,
help='the sample batch size')
parser.add_argument('--max_grad_norm', type=float, default=100)
parser.add_argument('--max_global_step', type=int, default=int(2e6))
parser.add_argument('--gpu', type=int, default=None)
# sac
parser.add_argument('--reward_scale',
type=float,
default=1.0,
help='reward scale')
# ppo
parser.add_argument('--clip_param', type=float, default=0.2)
parser.add_argument('--value_loss_coeff', type=float, default=0.5)
parser.add_argument('--action_loss_coeff', type=float, default=1.0)
parser.add_argument('--entropy_loss_coeff', type=float, default=1e-4)
parser.add_argument('--rollout_length', type=int, default=1000)
parser.add_argument('--gae_lambda', type=float, default=0.95)
# log
parser.add_argument('--log_interval', type=int, default=1)
parser.add_argument('--evaluate_interval', type=int, default=10)
parser.add_argument('--ckpt_interval', type=int, default=200)
parser.add_argument('--log_root_dir', type=str, default='log')
parser.add_argument('--wandb',
type=str2bool,
default=False,
help='set it True if you want to use wandb')
# evaluation
parser.add_argument('--ckpt_num', type=int, default=None)
parser.add_argument('--num_eval', type=int, default=10)
parser.add_argument('--save_rollout',
type=str2bool,
default=False,
help='save rollout information during evaluation')
parser.add_argument('--record',
type=str2bool,
default=True,
help='enable video recording')
parser.add_argument('--record_caption', type=str2bool, default=True)
parser.add_argument('--num_record_samples',
type=int,
default=1,
help='number of trajectories to collect during eval')
parser.add_argument(
'--save_qpos',
type=str2bool,
default=False,
help=
'save entire qpos history of success rollouts to file (for idle primitive training)'
)
parser.add_argument(
'--save_success_qpos',
type=str2bool,
default=True,
help=
'save later segment of success rollouts to file (for moving and placing primitie trainings)'
)
# misc
parser.add_argument('--prefix', type=str, default='test')
parser.add_argument('--notes', type=str, default='')
parser.add_argument('--seed', type=int, default=123, help='Random seed')
parser.add_argument('--debug', type=str2bool, default=False)
args, unparsed = parser.parse_known_args()
args.env_args_str = args.env_args
return args, unparsed
