Online switching control with stability and regret guarantees

Yingying Li, James A Preiss, Na Li, Yiheng Lin, Adam Wierman, Jeff S Shamma
Proceedings of The 5th Annual Learning for Dynamics and Control Conference, PMLR 211:1138-1151, 2023.

Abstract

This paper considers online switching control with a finite candidate controller pool, an unknown dynamical system, and unknown cost functions. The candidate controllers can be unstabilizing policies. We only require at least one candidate controller to satisfy certain stability properties, but we do not know which one is stabilizing. We design an online algorithm that guarantees finite-gain stability throughout the duration of its execution. We also provide a sublinear policy regret guarantee compared with the optimal stabilizing candidate controller. Lastly, we numerically test our algorithm on quadrotor planar flights and compare it with a classical switching control algorithm, falsification-based switching, and a classical multi-armed bandit algorithm, Exp3 with batches.

Cite this Paper

BibTeX
@InProceedings{pmlr-v211-li23a, title = {Online switching control with stability and regret guarantees}, author = {Li, Yingying and Preiss, James A and Li, Na and Lin, Yiheng and Wierman, Adam and Shamma, Jeff S}, booktitle = {Proceedings of The 5th Annual Learning for Dynamics and Control Conference}, pages = {1138--1151}, year = {2023}, editor = {Matni, Nikolai and Morari, Manfred and Pappas, George J.}, volume = {211}, series = {Proceedings of Machine Learning Research}, month = {15--16 Jun}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v211/li23a/li23a.pdf}, url = {https://proceedings.mlr.press/v211/li23a.html}, abstract = {This paper considers online switching control with a finite candidate controller pool, an unknown dynamical system, and unknown cost functions. The candidate controllers can be unstabilizing policies. We only require at least one candidate controller to satisfy certain stability properties, but we do not know which one is stabilizing. We design an online algorithm that guarantees finite-gain stability throughout the duration of its execution. We also provide a sublinear policy regret guarantee compared with the optimal stabilizing candidate controller. Lastly, we numerically test our algorithm on quadrotor planar flights and compare it with a classical switching control algorithm, falsification-based switching, and a classical multi-armed bandit algorithm, Exp3 with batches.} }
Endnote
%0 Conference Paper %T Online switching control with stability and regret guarantees %A Yingying Li %A James A Preiss %A Na Li %A Yiheng Lin %A Adam Wierman %A Jeff S Shamma %B Proceedings of The 5th Annual Learning for Dynamics and Control Conference %C Proceedings of Machine Learning Research %D 2023 %E Nikolai Matni %E Manfred Morari %E George J. Pappas %F pmlr-v211-li23a %I PMLR %P 1138--1151 %U https://proceedings.mlr.press/v211/li23a.html %V 211 %X This paper considers online switching control with a finite candidate controller pool, an unknown dynamical system, and unknown cost functions. The candidate controllers can be unstabilizing policies. We only require at least one candidate controller to satisfy certain stability properties, but we do not know which one is stabilizing. We design an online algorithm that guarantees finite-gain stability throughout the duration of its execution. We also provide a sublinear policy regret guarantee compared with the optimal stabilizing candidate controller. Lastly, we numerically test our algorithm on quadrotor planar flights and compare it with a classical switching control algorithm, falsification-based switching, and a classical multi-armed bandit algorithm, Exp3 with batches.
APA
Li, Y., Preiss, J.A., Li, N., Lin, Y., Wierman, A. & Shamma, J.S.. (2023). Online switching control with stability and regret guarantees. Proceedings of The 5th Annual Learning for Dynamics and Control Conference, in Proceedings of Machine Learning Research 211:1138-1151 Available from https://proceedings.mlr.press/v211/li23a.html.

Related Material