Shaifalee Saxena

Shaifalee Saxena

Ph.D. student at UNM and Graduate Research Assistant at Los Alamos National Laboratory working on reinforcement learning, robotics, and control.

Deep Reinforcement Learning for Robotic Manipulation under Distribution Shift with Bounded Extremum Seeking

Published in arXiv preprint, 2026

This paper investigates a hybrid controller for robotic manipulation under out-of-distribution test conditions. The reinforcement learning policy provides fast manipulation behavior, while bounded extremum seeking adapts online when goals, contact conditions, or dynamics shift.

Status: Preprint.

Recommended citation: Shaifalee Saxena, Rafael Fierro, and Alexander Scheinker. (2026). "Deep Reinforcement Learning for Robotic Manipulation under Distribution Shift with Bounded Extremum Seeking." arXiv preprint arXiv:2604.01142.
Download Paper