Minimax Q-learning Control for Linear Systems Using the Wasserstein Metric

DOI10.1016/J.AUTOMATICA.2022.110850MaRDI QIDQ6351242arXivFDO

Authors Feiran Zhao, Keyou You

Publication date 13 October 2020

Published in arXiv

Full work available at URL https://doi.org/10.1016/J.AUTOMATICA.2022.110850

Mathematics Subject Classification ID

Optimal stochastic control (93E20)

arXiv classification

cs.SY eess.SY math.OC

Abstract: Stochastic optimal control usually requires an explicit dynamical model with probability distributions, which are difficult to obtain in practice. In this work, we consider the linear quadratic regulator (LQR) problem of unknown linear systems and adopt a Wasserstein penalty to address the distribution uncertainty of additive stochastic disturbances. By constructing an equivalent deterministic game of the penalized LQR problem, we propose a Q-learning method with convergence guarantees to learn an optimal minimax controller.

This page was built for publication: Minimax Q-learning Control for Linear Systems Using the Wasserstein Metric

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6351242)