Welcome to ksim, a modular and easy-to-use framework for training policies in simulation.
To install the framework:
pip install ksim
Make sure to install JAX correctly for your hardware (CPU or GPU). We recommend using conda rather than uv to avoid compatibility issues with MuJoCo on macOS.
We suggest getting started by following the examples.
Many of the design decisions in ksim are heavily influenced other reinforcement learning training libraries like Mujoco Playground, Isaac Lab and Maniskill. In particular, we are very grateful to the Mujoco and MJX maintainers for building a great, cross-platform simulator, and we hope ksim will help contribute to making it the ecosystem standard.
