A few weeks ago, Konstantinos Chatzilygeroudis visited Chalmers University of Technology, invited by Shivesh Kumar, to give a lecture and hands-on tutorial on Reinforcement Learning for Robotics for the course Athletic Intelligence in Robotics (AIR).
The AIR course focuses on the fundamentals of athletic intelligence in robotics, with emphasis on dynamic robots, robust sensorimotor control, and practical tutorials using Python implementations, robot simulations, and/or hardware experiments. The course covers topics such as underactuated systems, dynamical systems, optimization, control, reinforcement learning, and Sim2Real transfer.
The tutorial introduced students to reinforcement learning for legged robotics through a practical example: learning to walk with a quadruped in a few minutes. The tutorial material is publicly available on GitHub: https://github.com/air-course/tutorial14.
We thank Shivesh Kumar and the AIR students for the invitation, hospitality, and engaging discussions on reinforcement learning, robot control, and the challenges of deploying learning-based methods on real robotic systems.