DeepRacer Gym¶

DeepRacer

A Gymnasium wrapper for the AWS DeepRacer simulator. Each deepracer-v0 environment automatically launches its own containerized simulator (Docker/Podman/Apptainer), then exposes it through the standard gymnasium API.

Quick Example¶

import gymnasium as gym
import deepracer

env = gym.make("deepracer-v0")      # starts a simulator service on demand

observation, info = env.reset()

observation, reward, terminated, truncated, info = env.step(
    env.action_space.sample()
)

env.close()                         # stops + removes the simulator service

Getting Started

Install the package and run your first rollout.
Configuring Environments

Reward functions, action/observation spaces, tracks.
Gymnasium Interface

Observation space, action space, and terminal states.
Managing Environments

Caching, cleanup, resource limits, and parallel rollouts.
Reference

Config keys, the full track catalog, and env vars.

Citation¶

If you use DeepRacer Gym in your work, please cite it. You can use the following BibTeX entry:

@software{akbar_deepracer,
    author    = {Akbar, Uzair},
    title     = {{DeepRacer Gym}},
    year      = {2026},
    version   = {0.1.1},
    publisher = {Zenodo},
    doi       = {10.5281/zenodo.21360352},
    url       = {https://doi.org/10.5281/zenodo.21360352}
}

Or in plain text:

Akbar, U. (2026). DeepRacer Gym (Version 0.1.1) [Computer software]. Zenodo. https://doi.org/10.5281/zenodo.21360352

Note

This is the concept DOI, which always resolves to the latest release. To cite a specific version instead, use that release's version DOI from its Zenodo record.

References¶

Balaji, B. et al. (2020). DeepRacer: Autonomous Racing Platform for Experimentation with Sim2Real Reinforcement Learning. IEEE International Conference on Robotics and Automation (ICRA). https://doi.org/10.1109/ICRA40945.2020.9197465
AWS DeepRacer Community. (2026). deepracer-simapp (Version 6.0.5) [Computer software]. GitHub. https://github.com/aws-deepracer-community/deepracer-simapp

Acknowledgements¶

This project was originally developed for use in Georgia Tech's CS 7642 Reinforcement Learning class. Many thanks to the course staff for their support in sharing it with the broader community.