Title:Closed-loop Performance Optimization of Model Predictive Control with Robustness Guarantees

Abstract:Model mismatch and process noise are two frequently occurring phenomena that can drastically affect the performance of model predictive control (MPC) in practical applications. We propose a principled way to tune the cost function and the constraints of linear MPC schemes to improve the closed-loop performance and robust constraint satisfaction on uncertain nonlinear dynamics with additive noise. The tuning is performed using a novel MPC tuning algorithm based on backpropagation developed in our earlier work. Using the scenario approach, we provide probabilistic bounds on the likelihood of closed-loop constraint violation over a finite horizon. We showcase the effectiveness of the proposed method on linear and nonlinear simulation examples.