Announcement: KUPAC 4th Physical AI Hands-on Workshop

This hands-on workshop uses the Unitree Go2 quadruped to explore reinforcement learning for robot walking, from the fundamentals through operation on a real robot.

Using the mjlab robot learning environment, participants will train a Go2 walking policy in simulation. They will work through environment setup, observation, action and reward design, then evaluate the learned policy.

Participants will then transfer the policy from simulation to a real Unitree Go2 and see the robot walk. By experiencing both simulation and hardware, the workshop aims to show how reinforcement learning is used to generate robot motion.