← Catalog
SimulatorGoogle DeepMind · 2025
MuJoCo Playground
Open-source suite of 50+ GPU-accelerated robot-learning environments built on MuJoCo MJX for sim-to-real research.
Summary
MuJoCo Playground is a GPU-accelerated library of 50+ environments spanning classic control, locomotion, and manipulation, built on MuJoCo MJX (with a MuJoCo Warp backend) for massively-parallel training and sim-to-real transfer. Winner of the RSS 2025 Outstanding Demo Paper award. Apache-2.0.
Metadata
- Organization
- Google DeepMind
- Year
- 2025
- License
- Apache-2.0
- Built on
- MuJoCo MJX (+ MuJoCo Warp)
- Environments
- 50+ (control, locomotion, manipulation)
Relationships
Extends
Compatible with
#rl#sim-to-real#mjx