mtrl.env package¶

Subpackages¶

mtrl.env.gym_1 package

Submodules¶

mtrl.env.builder module¶

mtrl.env.builder.build_dmcontrol_vec_env(domain_name: str, task_name: str, prefix: str, make_kwargs: omegaconf.dictconfig.DictConfig, env_id_list: List[int], seed_list: List[int], mode_list: List[str]) → mtrl.env.vec_env.VecEnv [source]¶

mtrl.env.builder.build_metaworld_vec_env(config: omegaconf.dictconfig.DictConfig, benchmark: metaworld.Benchmark, mode: str, env_id_to_task_map: Optional[Dict[str, metaworld.Task]]) → Tuple[gym.vector.async_vector_env.AsyncVectorEnv, Optional[Dict[str, Any]]][source]¶

mtrl.env.types module¶

Collection of types used in the env.

mtrl.env.vec_env module¶

class mtrl.env.vec_env.MetaWorldVecEnv(env_metadata: Dict[str, Any], env_fns, observation_space=None, action_space=None, shared_memory=True, copy=True, context=None, daemon=True, worker=None)[source]¶

Bases: gym.vector.async_vector_env.AsyncVectorEnv

Return only every skip-th frame

create_multitask_obs(env_obs)[source]¶

property ids¶

property mode¶

reset()[source]¶

Reset all sub-environments and return a batch of initial observations.

Returns: observations – A batch of observations from the vectorized environment.
Return type: sample from observation_space

step(actions)[source]¶

Take an action for each sub-environments.

Parameters

actions (iterable of samples from action_space) – List of actions.

Returns

observations (sample from observation_space) – A batch of observations from the vectorized environment.
rewards (np.ndarray instance (dtype np.float_)) – A vector of rewards from the vectorized environment.
dones (np.ndarray instance (dtype np.bool_)) – A vector whose entries indicate whether the episode has ended.
infos (list of dict) – A list of auxiliary diagnostic information dicts from sub-environments.

class mtrl.env.vec_env.VecEnv(env_metadata: Dict[str, Any], env_fns, observation_space=None, action_space=None, shared_memory=True, copy=True, context=None, daemon=True, worker=None)[source]¶