popgym.wrappers.previous_action

Module Contents

Classes

PreviousAction

Wrapper that adds the last action to the observation.

Attributes

PREV_ACTION

popgym.wrappers.previous_action.PREV_ACTION = 'prev_action'

class popgym.wrappers.previous_action.PreviousAction(env: popgym.core.env.POPGymEnv, null_action: gymnasium.core.ActType | None = None)

Bases: popgym.core.wrapper.POPGymWrapper

Wrapper that adds the last action to the observation.

Args:

env: The environment null_action: Optional null action that is returned when resetting the

environment. If not provided, the null action will be 0 (int or vector) if it is in the action space, or the lowest action possible.

Returns:

A gym environment

static add_act_space_to_obs_space(observation_space: gymnasium.spaces.Space, action_space: gymnasium.spaces.Space) → gymnasium.spaces.Space

Returns a modified observation space to account for the last action. Args:

observation_space: Original observation space action_space: Action space

Returns:

The new observation space

static add_act_to_obs(observation_space: gymnasium.spaces.Space, obs: gymnasium.core.ObsType, action: gymnasium.core.ActType) → gymnasium.core.ObsType

Static method that adds the action to the observation. Args:

observation_space: Original observation space of the environment. obs: The observation. action: The action.

Returns:

Modified observation.

static get_null_action(action_space: gymnasium.spaces.Space) → gymnasium.core.ActType

Static method that generates a null action based on the action space. Args:

action_space: The action space.

Returns:

The null action.

step(action: gymnasium.core.ActType) → Tuple[gymnasium.core.ObsType, float, bool, bool, dict]

Uses the step() of the env that can be overwritten to change the returned data.

reset(**kwargs)

Uses the reset() of the env that can be overwritten to change the returned data.