Restriction: Must be in the Computer Science Master's or Doctoral program; or permission of instructor

This course introduces core methods for enabling robots to learn from data. We will focus on visuomotor policy learning for manipulation, covering topics such as behavior cloning, offline/online RL, sim-to-realtransfer, and multimodal perception (vision, touch, force). We will alsodiscuss challenges in robotics such as embodiment, data efficiency, and generalization across tasks and platforms, along with emerging directions like foundation models and world models. This is a seminar-style course based on recent research papers, with student presentations and a final project.