Towards a Robotic Hand Rehabilitation Exoskeleton for Stroke Therapy

Abstract

A majority of stroke patients suffer from the loss of effective motor function, which compromises their ability to control grasping motion. Hand rehabilitation is therefore important to improve their motor function and quality of life in activities of daily living (ADLs). In this initial work, we present the design and development of a partial hand exoskeleton actuated by shape memory alloy (SMA) spring actuators. The SMA spring actuators are cooled by forced convection and the individual joints of the finger are actuated via tendons. In this design, pre-tension in the passive springs enables the restoration of the original configuration when the SMA springs are not actuated. To address the slow cooling rate of SMA springs that limits the control performance, we developed a cooling unit for each SMA spring actuator. We utilized computer vision to identify an object and provide 3-D coordinates of the optimal grasping points on the object. We then utilized vision-based control to move the fingertips towards the grasping points. The experimental results showed that each individual joint was able to return to its original configuration significantly faster as well as to follow a sinusoidal trajectory with the proposed cooling strategy.