Dynamic Regret Minimization for Control of Non-stationary Linear Dynamical Systems
Abstract
We consider the problem of controlling a Linear Quadratic Regulator (LQR) system over a finite horizon with fixed and known cost matrices , but unknown and non-stationary dynamics . The sequence of dynamics matrices can be arbitrary, but with a total variation, , assumed to be and unknown to the controller. Under the assumption that a sequence of stabilizing, but potentially sub-optimal controllers is available for all , we present an algorithm that achieves the optimal dynamic regret of . With piece-wise constant dynamics, our algorithm achieves the optimal regret of where is the number of switches. The crux of our algorithm is an adaptive non-stationarity detection strategy, which builds on an approach recently developed for contextual Multi-armed Bandit problems. We also argue that non-adaptive forgetting (e.g., restarting or using sliding window learning with a static window size) may not be regret optimal for the LQR problem, even when the window size is optimally tuned with the knowledge of . The main technical challenge in the analysis of our algorithm is to prove that the ordinary least squares (OLS) estimator has a small bias when the parameter to be estimated is non-stationary. Our analysis also highlights that the key motif driving the regret is that the LQR problem is in spirit a bandit problem with linear feedback and locally quadratic cost. This motif is more universal than the LQR problem itself, and therefore we believe our results should find wider application.
Cite
@article{arxiv.2111.03772,
title = {Dynamic Regret Minimization for Control of Non-stationary Linear Dynamical Systems},
author = {Yuwei Luo and Varun Gupta and Mladen Kolar},
journal= {arXiv preprint arXiv:2111.03772},
year = {2022}
}