sinergym.utils.wrappers.MultiObsWrapper

class sinergym.utils.wrappers.MultiObsWrapper(env: Any, n: int = 5, flatten: bool = True)

__init__(env: Any, n: int = 5, flatten: bool = True) → None

Stack of observations.

Parameters

• env (Any) – Original Gym environment.

• n (int, optional) – Number of observations to be stacked. Defaults to 5.

• flatten (bool, optional) – Whether or not flat the observation vector. Defaults to True.

Methods

__init__(env[, n, flatten])	Stack of observations.
class_name()
close()	Override close in your subclass to perform any necessary cleanup.
compute_reward(achieved_goal, desired_goal, info)
render([mode])	Renders the environment.
reset()	Resets the environment.
seed([seed])	Sets the seed for this env's random number generator(s).
step(action)	Performs the action in the new environment.

Attributes

action_space
metadata
observation_space
reward_range
spec
unwrapped	Completely unwrap this env.

reset() → numpy.ndarray

Resets the environment.

Returns

Stacked previous observations.

Return type

np.ndarray

step(action: Union[int, numpy.ndarray]) → Tuple[numpy.ndarray, float, bool, Dict[str, Any]]

Performs the action in the new environment.

Parameters

action (Union[int, np.ndarray]) – Action to be executed in environment.

Returns

Tuple with next observation, reward, bool for terminated episode and dict with extra information.

Return type

Tuple[np.ndarray, float, bool, Dict[str, Any]]