3.0

Average Course Rating

Model-based methods provide a powerful framework for controlling challenging robotic systems; however, imperfect models often lead to poor performance during real-world deployment. Machine learning methods provide one means of addressing deficient models, either through explicitly learning a model of the system dynamics or computing a control policy directly from data. In this course, we will explore the intersection between optimal control and machine learning, covering both model-free and model-based methods for learning-based control. We will start with a review of dynamic programming and its relationship to reinforcement learning. We will then explore the three primary means of incorporating learning into controller synthesis: learning value functions, control policies, and dynamics models. The course will culminate in a discussion of model-based reinforcement learning and adaptive optimal control. We will also discuss advanced topics such as learning Lyapunov functions and contraction metrics from data, iterative learning control, and techniques for adaptive nonlinear model predictive control. No Audits allowed. Recommended Course Background: Course work in1) differential equations 2) multi-variable calculus 3) linear algebra 4) undergraduate linear control 5) graduate-level introductory robotics course such as EN.530.646 Robot Devices, Kinematics, Dynamics and Control and 5) MATLAB and/or Python programming.