NIDILRR ARRT Fellows Symposium: A Model-free Reinforcement Learning-based Control Approach to Provide Personalized Assistance for People with Stroke

Powered exoskeletons are promising devices to improve the walking patterns of people with neurological impairments. However, providing optimal personalized external assistance is challenging due to modeling uncertainties and time-varying human-robot interaction. We propose a model-free reinforcement learning (RL)-based hierarchical control framework to provide adaptive and optimal personalized exoskeleton assistance, to achieve a normative range of motion (ROM) on the affected hip joint for individuals with stroke during walking.
Source: Archives of Physical Medicine and Rehabilitation - Category: Rehabilitation Authors: Tags: NIDILRR ARRT 2420347 Source Type: research