This graduate-level course delves into the intersection of machine learning and robotics. The curriculum will explore the integration of contemporary learning techniques in robotic areas such as manipulation, navigation, planning, control, decision-making, and other pertinent challenges in robotics. Advanced deep learning techniques and their applications in robotics will be covered, including supervised learning (e.g., behavioral cloning, state prediction), reinforcement learning (e.g., actor-critic, visual foresight), and unsupervised/self-supervised methods (e.g., world model construction, learning forward dynamic models). In addition, the generalizability of these methods will be discussed, recent, and experimental studies will be conducted, examining the challenges of applying these techniques on physical systems.
RBE 500 or equivalent
RBE 501 and RBE 502.