Adaptive Control of a Wearable Exoskeleton for Upper-Extremity Neurorehabilitation

Abstract

The paper describes the implementation and testing of two adaptive controllers developed for a wearable, underactuated upper extremity therapy robot – RUPERT (Robotic Upper Extremity Repetitive Trainer). The controllers developed in this study were used to implement two adaptive robotic therapy modes – the adaptive co-operative mode and the adaptive active-assist mode – that are based on two different approaches for providing robotic assistance for task practice. The adaptive active-assist mode completes therapy tasks when a subject is unable to do so voluntarily. This robotic therapy mode is a novel implementation of the idea of an active-assist therapy mode; it utilizes the measure of a subject’s motor ability, along with their real-time movement kinematics to initiate robotic assistance at the appropriate time during a movement trial. The adaptive co-operative mode, on the other hand, is based on the idea of enabling task completion instead of completing the task for the subject. Both these therapy modes were designed to adapt to a stroke subject's motor ability, and thus encourage voluntary participation from the stroke subject. The two controllers were tested on three stroke subjects practicing robot-assisted reaching movements. The results from this testing demonstrate that an underactuated wearable exoskeleton, such as RUPERT, can be used for administering robot-assisted therapy, in a manner that encourages voluntary participation from the subject undergoing therapy.