Download
MuJoCo Playground, developed by Google DeepMind, is a GPU-accelerated suite of simulation environments for robot learning and sim-to-real research, built on top of MuJoCo MJX. It unifies a range of control, locomotion, and manipulation tasks into a consistent and scalable framework optimized for JAX and Warp backends. The project includes classic control benchmarks from dm_control, advanced quadruped and bipedal locomotion systems, and dexterous as well as non-prehensile manipulation setups. It also offers optional vision-based training capabilities through integration with Madrona-MJX, allowing researchers to train policies directly from image input on GPUs. MuJoCo Playground supports both the MJX JAX implementation and the Warp physics engine, enabling flexible use across research pipelines. The environments are designed for fast training, compatibility with reinforcement learning libraries, and real-time trajectory visualization using rscope.
Features
- GPU-accelerated physics simulation via MuJoCo MJX and Warp backends
- Wide range of environments for control, locomotion, and manipulation
- Vision-based learning support through Madrona-MJX integration
- JAX-compatible training pipelines with example PPO scripts
- Interactive trajectory visualization via rscope
- Reproducible, research-grade environments with CUDA and Colab support
Project Samples
