Improved Robustness of Deep Reinforcement Learning for Control of Time-Varying Systems by Bounded Extremum Seeking

This paper studies how bounded extremum seeking can improve the robustness of deep reinforcement learning controllers when system dynamics change over time. The approach combines fast learned policies with online adaptation to better handle nonlinear and time-varying behavior.

Status: Accepted to ACC 2026.

Recommended citation: Shaifalee Saxena, Alan Williams, Rafael Fierro, and Alexander Scheinker. (2026). "Improved Robustness of Deep Reinforcement Learning for Control of Time-Varying Systems by Bounded Extremum Seeking." American Control Conference (ACC 2026).
Download Paper