Time: | February 16, 2024 |
---|---|
Download as iCal: |
|
Dr. Jared F. Miller
Automatic Control Lab
ETH Zurich
Zurich, Switzerland
Friday 2024-02-16 2 p.m.
IST Seminar Room 2.255 - Pfaffenwaldring 9 - Campus Stuttgart-Vaihingen
Abstract
The Errors-in-Variables (EIV) setting involves the simultaneous presence of input and measurement noise, and induces nonconvex system identification and control problems. This presentation begins by performing set-membership direct data-driven control in the EIV setting by casting the original nonconvex stabilization problem as an infinite-dimensional linear program (restricting to fixed polyhedral Lyapunov functions). The unknown-but-bounded input and measurement noise terms can be eliminated using robust optimization, resulting in dual variables that are parameterized by the state-space system model. Such a robust decomposition reduces the computational complexity of discretizing the infinite-dimensional linear programs and increases tractability. This method of parameterized robust counterparts is then applied to optimal control problems with input-affine dynamics and semidefinite representable uncertainty. The optimal control problems can be further extended to data-driven safety quantification analysis problems (e.g. estimation of peak values, distances, reachable sets) in the continuous-time and discrete-time settings.
Biographical Information
Jared Miller is a postdoctoral researcher at the Automatic Control Lab, ETH Zurich, in the research group of Roy Smith. He received his B.S. and M.S. degrees in Electrical Engineering from Northeastern University in 2018 and his Ph.D. in Electrical Engineering from Northeastern University in 2023. He is a recipient of the 2020 Chateaubriand Fellowship from the Office for Science Technology of the Embassy of France in the United States. He was given an Outstanding Student Paper award at the IEEE Conference on Decision and Control in 2021 and in 2022, and was a finalist for the Young Author Award at the 2023 IFAC world congress. His current research topics include safety verification and data-driven control. His interests include large-scale convex optimization, nonlinear systems, semi-algebraic geometry, and measure theory.