About this project
MuJoCo (Multi-Joint dynamics with Contact) is a physics engine maintained by Google DeepMind, primarily used for research and development in robotics, biomechanics, machine learning, graphics, and animation.
Key technical features include:
- A high-performance runtime simulation module with a C API.
- A built-in XML compiler for preallocating low-level data structures.
- Native interactive visualization via an OpenGL-rendered GUI.
- Extensive utility functions for computing physics-related quantities.
Integration and accessibility options:
- Official bindings for Python, JavaScript (WebAssembly), and C# (Unity plug-in).
- Support for various third-party bindings including MATLAB Simulink, Swift, Java, Julia, and Rust.
- Compatibility with external model formats through converters for OpenSim, SDFormat, OBJ, and Onshape.
- MuJoCo XLA (MJX), a version written in JAX for differentiable physics and accelerated simulation.