Dynamic Potential Field-based Assisted-as-Needed Control Strategy for Robotic Post-Stroke Rehabilitation

Mahboobe Shakeri Nadrabadi, Jay Carriere

Keywords: Robot-assisted Rehabilitation, Assisted-as-Needed Control, Potential Field

Abstract: Stroke is one of the leading causes of disability among adults, and robot-assisted rehabilitation offers consistent, quantifiable, and intensive therapy. Assisted-as-needed (AAN) control is a key element of robot-assisted rehabilitation, as it minimizes robotic intervention while encouraging patients to contribute actively. However, designing effective AAN strategies for complex rehabilitation tasks remains a challenge. In this work, we propose a dynamic potential field–based AAN control strategy for a 2D rehabilitation task where patients navigate from a start point to an endpoint while avoiding random obstacles. The strategy combines a radial component, modeled with a sigmoid function to keep patients close to the pre-computed path, and a progression component, which promotes forward task completion. The proposed potential field is tunable both offline (safe radius, sharpness, initial progression, desired task speed) and online (progression rate), enabling individualized assistance based on patient history and real-time performance.

Submission Number: 30