Upcoming Event: Joint ASE/Oden Institute Seminar Series

Machine learning for improved dynamical models and certified control

Alexander (Sasha) Davydov, Rice University

Abstract

For several classes of nonlinear systems, we have neither good first-principles models nor good approaches for controlling them. Thus, it is becoming increasingly common to use machine learning approaches to model them and control them. In this talk, I will present two recent works on leveraging machine learning for systems modeling and certified control. In the first part of the talk, I will discuss how we can improve learned models using Bayesian meta-learning and active information gathering and how we have applied this methodology in driving at the limits of handling. In the second part of the talk, I will discuss learning controllers that provably provide closed-loop contraction guarantees. The training methodology is based on interval analysis and the implementation is aided by advances in automatic differentiation of interval-based methods.

Biography

Alexander (Sasha) Davydov is an assistant professor in the Department of Mechanical Engineering at Rice University where he joined in Fall 2025. Prior to Rice University, he was a Ph.D. student at UC Santa Barbara in the Department of Mechanical Engineering and Center for Control, Dynamical Systems, and Computation. He is the recipient of several awards including the 2024 IEEE Control Systems Letters Outstanding Paper Award, the 2023 AACC O. Hugo Schuck Best Paper Award, and the 2025 UCSB Department of Mechanical Engineering, Best Ph.D. Thesis award. His research interests are broadly at the intersection of learning and optimization for control with applications of interest in robotic systems and autonomous driving.