X2 Sim + MC Deployment

Note: This document targets x86 host-side simulation, driving the MuJoCo simulation sim_mujoco with the motion-control module mc; robot model lx2501_3_t2d5 (X2 T2.5). Together with RL Deployment these are two parallel simulation pipelines — pick one; do not drive the same simulation at the same time.

  • sim_mujoco: MuJoCo-based physics simulation acting as the robot body (HAL layer) — subscribes to joint commands and publishes joint and IMU states

  • mc: motion-control module, identical to the real robot, with built-in skills (standing / walking / getting up / animations, etc.) — receives high-level commands and outputs joint commands

  • Drive the robot by sending high-level commands via the SDK examples (py_examples / examples); suitable for competition simulation and host-side algorithm integration.

Note

When to use this chapter: no reinforcement learning involved — drive the MuJoCo simulation with the official motion-control module mc (the same one used on the real robot) and validate your own high-level control logic on an x86 host. The skills (stand / walk / get-up / animation / upper body) are provided by mc, so you do not need to train a policy yourself.

If you only want to wire an existing onnx model into the SDK, see X2 RL Deployment.