MuJoCo MJX

Learn robot simulation with MuJoCo and scale it with JAX, MJX, vectorized rollouts, domain randomization, and Playground PPO.

Goals

• Understand MJCF, contacts, cameras, control, and inverse kinematics
• Move MuJoCo models into MJX
• Use JIT and vmap for parallel rollouts
• Train Playground PPO tasks including Franka pick-cube

Foundations

• MJX01 — Concepts and MJCF

  Understand MuJoCo concepts and model robots with MJCF.

• MJX02 — Rendering, Cameras, Contacts

  Render scenes and inspect cameras and contact dynamics.

• MJX03 — Control and IK

  Control robot arms with damped least-squares inverse kinematics.

• MJX04 — From MuJoCo to MJX

  Move a simulation into JAX-accelerated MJX.

• MJX05 — Parallel Rollouts

  Vectorize rollouts and apply domain randomization.

Playground Reinforcement Learning

• MJX06 — RL Refresher

  Review MDP and PPO concepts for continuous control.

• MJX07 — Playground PPO

  Train a PPO control task in MuJoCo Playground.

• MJX08 — PointMass

  Train a PointMass navigation policy.

• MJX09 — Panda Pick Cube

  Train a Franka Panda pick-cube policy.

Return to the Physical AI overview.