Almost Surely $\sqrt{T}$ Regret for Adaptive LQR

TL;DR

This paper introduces an adaptive LQR controller that achieves almost sure (rac{ T}) regret bounds, ensuring optimal long-term performance with safety guarantees, validated through industrial process simulations.

Contribution

It presents a novel adaptive LQR controller with almost sure (rac{ T}) regret, incorporating a circuit-breaking mechanism for safety and convergence.

Findings

Achieves (rac{ T}) regret bounds almost surely.

Circuit-breaking mechanism ensures safety and finite triggers.

Validated on Tennessee Eastman Process simulation.

Abstract

The Linear-Quadratic Regulation (LQR) problem with unknown system parameters has been widely studied, but it has remained unclear whether $\tilde{ \mathcal{O}}(\sqrt{T})$ regret, which is the best known dependence on time, can be achieved almost surely. In this paper, we propose an adaptive LQR controller with almost surely $\tilde{ \mathcal{O}}(\sqrt{T})$ regret upper bound. The controller features a circuit-breaking mechanism, which circumvents potential safety breach and guarantees the convergence of the system parameter estimate, but is shown to be triggered only finitely often and hence has negligible effect on the asymptotic performance of the controller. The proposed controller is also validated via simulation on Tennessee Eastman Process~(TEP), a commonly used industrial process example.