Explores the use of data to design safe and effective controllers, through the lens of model predictive control. Introduction to standard optimal control and model predictive control formulations, with emphasis on implementation rather than theory. Introduction to “learning model predictive control”, which then serves as a foundation for evaluating performance and safety of other common data-driven control methods (e.g. Reinforcement Learning and Imitation Learning). The course is implementation-focused through a combination of lectures and simulation-based control design activities. Experience with controls is required; experience with optimization is helpful. Prereqs: 155A or equivalent.
3
UnitsLetter
Grading1, 2, 3
PasstimeNone
Level LimitEngineering
College