Weeks 13-14 – Starting reinfortment learning
During this two weeks what I’ve mainly done is traning both the G1 and the Go2 using reinfortment learning hoe to walk on a flat surface.
For the G1 robot, I used the example provided in the mjlab repository. This setup allowed me to train a basic locomotion policy using reinforcement learning and obtain the following results: G1 movement
For the Go2 robot, the training code was obtained from the official [unitree mjlab] repository. Using this framework, I trained the model to achieve basic locomotion behavior:(https://github.com/unitreerobotics/unitree_rl_mjlab) Go2 movement
Summary
- Trainig G1
- Training Go2