{"id":28753271,"url":"https://github.com/google-deepmind/mujoco_playground","last_synced_at":"2025-06-17T00:39:24.729Z","repository":{"id":272679789,"uuid":"898162230","full_name":"google-deepmind/mujoco_playground","owner":"google-deepmind","description":"An open-source library for GPU-accelerated robot learning and sim-to-real transfer.","archived":false,"fork":false,"pushed_at":"2025-05-21T18:59:19.000Z","size":309334,"stargazers_count":1023,"open_issues_count":17,"forks_count":120,"subscribers_count":17,"default_branch":"main","last_synced_at":"2025-05-21T19:53:10.980Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":"https://playground.mujoco.org","language":"Jupyter Notebook","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"apache-2.0","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/google-deepmind.png","metadata":{"files":{"readme":"README.md","changelog":"CHANGELOG.md","contributing":"CONTRIBUTING.md","funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":"CITATION.cff","codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null}},"created_at":"2024-12-03T22:39:09.000Z","updated_at":"2025-05-21T18:59:23.000Z","dependencies_parsed_at":"2025-01-16T00:56:19.508Z","dependency_job_id":"221635e9-1405-42de-9b44-8d875e32ce38","html_url":"https://github.com/google-deepmind/mujoco_playground","commit_stats":null,"previous_names":["google-deepmind/mujoco_playground"],"tags_count":3,"template":false,"template_full_name":null,"purl":"pkg:github/google-deepmind/mujoco_playground","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/google-deepmind%2Fmujoco_playground","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/google-deepmind%2Fmujoco_playground/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/google-deepmind%2Fmujoco_playground/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/google-deepmind%2Fmujoco_playground/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/google-deepmind","download_url":"https://codeload.github.com/google-deepmind/mujoco_playground/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/google-deepmind%2Fmujoco_playground/sbom","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":260268635,"owners_count":22983601,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2025-06-17T00:39:23.867Z","updated_at":"2025-06-17T00:39:24.695Z","avatar_url":"https://github.com/google-deepmind.png","language":"Jupyter Notebook","funding_links":[],"categories":["Simulation Benchmark"],"sub_categories":[],"readme":"# MuJoCo Playground\n\n[![Build](https://img.shields.io/github/actions/workflow/status/google-deepmind/mujoco_playground/ci.yml?branch=main)](https://github.com/google-deepmind/mujoco_playground/actions)\n[![PyPI version](https://img.shields.io/pypi/v/playground)](https://pypi.org/project/playground/)\n![Banner for playground](https://github.com/google-deepmind/mujoco_playground/blob/main/assets/banner.png?raw=true)\n\nA comprehensive suite of GPU-accelerated environments for robot learning research and sim-to-real, built with [MuJoCo MJX](https://github.com/google-deepmind/mujoco/tree/main/mjx).\n\nFeatures include:\n\n- Classic control environments from `dm_control` reimplemented in MJX.\n- Quadruped and bipedal locomotion environments.\n- Non-prehensile and dexterous manipulation environments.\n- Vision-based support available via [Madrona-MJX](https://github.com/shacklettbp/madrona_mjx).\n\nFor more details, check out the project [website](https://playground.mujoco.org/).\n\n## Installation\n\nYou can install MuJoCo Playground directly from PyPI:\n\n```sh\npip install playground\n```\n\n### From Source\n\n\u003e [!IMPORTANT]\n\u003e Requires Python 3.10 or later.\n\n1. `git clone git@github.com:google-deepmind/mujoco_playground.git \u0026\u0026 cd mujoco_playground`\n2. [Install uv](https://docs.astral.sh/uv/getting-started/installation/), a faster alternative to `pip`\n3. Create a virtual environment: `uv venv --python 3.11`\n4. Activate it: `source .venv/bin/activate`\n5. Install CUDA 12 jax: `uv pip install -U \"jax[cuda12]\"`\n    * Verify GPU backend: `python -c \"import jax; print(jax.default_backend())\"` should print gpu\n6. Install playground: `uv pip install -e \".[all]\"`\n7. Verify installation (and download Menagerie): `python -c \"import mujoco_playground\"`\n\n#### Madrona-MJX (optional)\n\nFor vision-based environments, please refer to the installation instructions in the [Madrona-MJX](https://github.com/shacklettbp/madrona_mjx?tab=readme-ov-file#installation) repository.\n\n## Getting started\n\n### Basic Tutorials\n| Colab | Description |\n|-------|-------------|\n| [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/google-deepmind/mujoco_playground/blob/main/learning/notebooks/dm_control_suite.ipynb) | Introduction to the Playground with DM Control Suite |\n| [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/google-deepmind/mujoco_playground/blob/main/learning/notebooks/locomotion.ipynb) | Locomotion Environments |\n| [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/google-deepmind/mujoco_playground/blob/main/learning/notebooks/manipulation.ipynb) | Manipulation Environments |\n\n### Vision-Based Tutorials (GPU Colab)\n| Colab | Description |\n|-------|-------------|\n| [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/google-deepmind/mujoco_playground/blob/main/learning/notebooks/training_vision_1_t4.ipynb) | Training CartPole from Vision (T4 Instance) |\n\n### Local Runtime Tutorials\n*Requires local Madrona-MJX installation*\n\n| Colab | Description |\n|-------|-------------|\n| [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/google-deepmind/mujoco_playground/blob/main/learning/notebooks/training_vision_1.ipynb) | Training CartPole from Vision |\n| [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/google-deepmind/mujoco_playground/blob/main/learning/notebooks/training_vision_2.ipynb) | Robotic Manipulation from Vision |\n\n## Running from CLI\n\u003e [!IMPORTANT]\n\u003e Assumes installation from source.\n\nFor basic usage, navigate to the repo's directory and run:\n```bash\npython learning/train_jax_ppo.py --env_name CartpoleBalance\n```\n\n### Training Visualization\n\nTo interactively view trajectories throughout training with [rscope](https://github.com/Andrew-Luo1/rscope/tree/main), install it (`pip install rscope`) and run:\n\n```\npython learning/train_jax_ppo.py --env_name PandaPickCube --rscope_envs 16 --run_evals=False --deterministic_rscope=True\n# In a separate terminal\npython -m rscope\n```\n\n## FAQ\n\n### How can I contribute?\n\nGet started by installing the library and exploring its features! Found a bug? Report it in the issue tracker. Interested in contributing? If you are a developer with robotics experience, we would love your help—check out the [contribution guidelines](CONTRIBUTING.md) for more details.\n\n### Reproducibility / GPU Precision Issues\nUsers with NVIDIA Ampere architecture GPUs (e.g., RTX 30 and 40 series) may experience reproducibility [issues](https://github.com/google-deepmind/mujoco_playground/issues/86) in mujoco_playground due to JAX’s default use of TF32 for matrix multiplications. This lower precision can adversely affect RL training stability. To ensure consistent behavior with systems using full float32 precision (as on Turing GPUs), please run `export JAX_DEFAULT_MATMUL_PRECISION=highest` in your terminal before starting your experiments (or add it to the end of `~/.bashrc`).\n\n## Citation\n\nIf you use Playground in your scientific works, please cite it as follows:\n\n```bibtex\n@misc{mujoco_playground_2025,\n  title = {MuJoCo Playground: An open-source framework for GPU-accelerated robot learning and sim-to-real transfer.},\n  author = {Zakka, Kevin and Tabanpour, Baruch and Liao, Qiayuan and Haiderbhai, Mustafa and Holt, Samuel and Luo, Jing Yuan and Allshire, Arthur and Frey, Erik and Sreenath, Koushil and Kahrs, Lueder A. and Sferrazza, Carlo and Tassa, Yuval and Abbeel, Pieter},\n  year = {2025},\n  publisher = {GitHub},\n  url = {https://github.com/google-deepmind/mujoco_playground}\n}\n```\n\n## License and Disclaimer\n\nThe texture used in the rough terrain for the locomotion environments is from [Polyhaven](https://polyhaven.com/a/rock_face) and licensed under [CC0](https://creativecommons.org/public-domain/cc0/).\n\nAll other content in this repository is licensed under the Apache License, Version 2.0. A copy of this license is provided in the top-level [LICENSE](LICENSE) file in this repository. You can also obtain it from https://www.apache.org/licenses/LICENSE-2.0.\n\nThis is not an officially supported Google product.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fgoogle-deepmind%2Fmujoco_playground","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fgoogle-deepmind%2Fmujoco_playground","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fgoogle-deepmind%2Fmujoco_playground/lists"}