Implementation:Farama Foundation Gymnasium Vector Action Wrappers

Knowledge Sources	Farama_Foundation_Gymnasium Gymnasium Docs
Domains	Reinforcement_Learning, Wrappers
Last Updated	2026-02-15 03:00 GMT

Overview

A collection of vectorized action wrappers that transform actions for VectorEnv environments, including TransformAction, VectorizeTransformAction, ClipAction, and RescaleAction.

Description

This module provides vectorized versions of action transformation wrappers:

• TransformAction -- Applies a user-provided function to the entire action vector at once. Accepts optional action_space and single_action_space parameters for specifying the transformed spaces. Best used when vector-level parallel processing of actions is possible.

• VectorizeTransformAction -- Wraps a single-agent action wrapper to work with vector environments by iterating over individual actions and applying the wrapper's transform. Uses a fake _SingleEnv internally. Serves as the base class for ClipAction and RescaleAction.

• ClipAction -- Clips continuous actions to the environment's action space bounds. A thin wrapper around VectorizeTransformAction using the single-agent ClipAction wrapper.

• RescaleAction -- Affinely rescales the continuous action space to a new range [min_action, max_action]. A thin wrapper around VectorizeTransformAction using the single-agent RescaleAction wrapper.

Usage

Use these wrappers to transform actions in vectorized environments. ClipAction is commonly used to ensure valid actions, and RescaleAction is used to normalize action ranges. Use TransformAction for custom vector-level action transforms.

Code Reference

Source Location

• Repository: Farama_Foundation_Gymnasium
• File: gymnasium/wrappers/vector/vectorize_action.py

Signature

class TransformAction(VectorActionWrapper):
    def __init__(self, env: VectorEnv, func: Callable[[ActType], Any],
                 action_space: Space | None = None, single_action_space: Space | None = None): ...

class VectorizeTransformAction(VectorActionWrapper):
    def __init__(self, env: VectorEnv, wrapper: type[transform_action.TransformAction], **kwargs: Any): ...

class ClipAction(VectorizeTransformAction):
    def __init__(self, env: VectorEnv): ...

class RescaleAction(VectorizeTransformAction):
    def __init__(self, env: VectorEnv, min_action: float | int | np.ndarray, max_action: float | int | np.ndarray): ...

Import

from gymnasium.wrappers.vector import TransformAction, VectorizeTransformAction, ClipAction, RescaleAction

I/O Contract

Inputs

Name	Type	Required	Description
env	VectorEnv	Yes	The vector environment to wrap
func	Callable	Yes (TransformAction)	Function to transform the action vector
action_space	Space or None	No	New action space for the wrapper
single_action_space	Space or None	No	Action space for a single sub-environment
min_action	float, int, or ndarray	Yes (RescaleAction)	Minimum values for rescaled action space
max_action	float, int, or ndarray	Yes (RescaleAction)	Maximum values for rescaled action space

Outputs

Name	Type	Description
actions	ActType	Transformed actions passed to the underlying vector environment

Usage Examples

import numpy as np
import gymnasium as gym
from gymnasium.wrappers.vector import ClipAction, RescaleAction

# ClipAction: clip out-of-bounds actions
envs = gym.make_vec("MountainCarContinuous-v0", num_envs=3)
envs = ClipAction(envs)
obs, info = envs.reset(seed=123)
obs, rew, term, trunc, info = envs.step(np.array([5.0, -5.0, 2.0]))
envs.close()

# RescaleAction: rescale actions to [0, 1]
envs = gym.make_vec("MountainCarContinuous-v0", num_envs=3)
envs = RescaleAction(envs, 0.0, 1.0)
_ = envs.action_space.seed(123)
obs, info = envs.reset(seed=123)
for _ in range(10):
    obs, rew, term, trunc, info = envs.step(0.5 * np.ones((3, 1)))
envs.close()

Related Pages

• Environment:Farama_Foundation_Gymnasium_Python_3_10_Runtime