Simulation with Mujoco

We have been working on a more unified simulation runner for GBC rather than running deployment directly provided by Unitree RL Gym. This part of the code will be uploaded once we are done (mapping joints and controllers from isaac lab configuration to mujoco automatically is a huge pile of work, and since I am currently busy applying for PhD, I have little time to do this now).

For users using the trained policies, we recommend using the RL repositories aligned with your type of robot, and use their sim-to-sim scripts. Just replace the onnx file and reformulate the observations & reference observations the way we implemented here. However, since our model is designed to work without reference, keeping the original repo's sim-to-sim script is fine.