Dominic Liao-McPherson
PhD Candidate
UM Aerospace Engineering

Model Predictive Control (MPC) is a powerful control methodology that constructs a control law from the solution of a receding horizon optimal control problem (OCP). MPC can systemically handle nonlinearities, coupling, and constraints but can be difficult to implement because of the need to solve non-linear OCPs online. One way to reduce this computational burden is to exploit that in MPC one solves a sequence of OCPs and reuse information from previous problems, a practice commonly called "warmstarting". In this talk, I discuss the theoretical, algorithmic, and practical application of warmstarting in MPC. First, I introduce Time-distributed Optimization (TDO), a unifying framework for studying the system theoretic consequence of warmstarting, which we use to derive sufficient conditions for stability and robustness. Second, I present FBstab, a quadratic programming algorithm with strong robustness properties that is designed to be warmstarted and can exploit the structure of optimal control problems. Finally, I illustrate the applicability of the these methods in the real-world, using diesel engine, autonomous driving, and guided parafoil examples.

About the Speaker...

Dominic Liao-McPherson obtained his BASc (with High Honours) in Engineering Science, Aerospace Option, from the University of Toronto in 2015. Since 2015 he has been a PhD student at the University of Michigan, in the department of aerospace engineering. His research interests lie in model predictive control, reference governors, trajectory optimization, and numerical methods with applications in aerospace, robotics, and autonomous vehicles. He received the 2019 Prof. Kabamba award and a predoctoral fellowship from the University of Michigan and was a finalist in the 2019 ECC best student paper competition.