Development of Hybrid-Actuator Robotic Exoskeleton Based on Gesture Signal Recognition Algorithm for the Rehabilitation of Dysfunctional Finger

Abstract

The present work, which describes the development of a novel, portable, low-cost, effective, hybrid-actuator rehabilitation exoskeleton, aims to present a solution for the rehabilitation of functional finger injuries. In this robotic system, a simple and ingenious actuator is designed on the synchronizing wheel of each finger joint, which enables the independent passive training of each finger joint with the actuation of the motor. In addition, three damping shafts with leaf springs as another type of actuator, corresponding to PIP, MIP and DIP joints, are used as damping devices to supply the damping force for active training. Moreover, a gesture-based signal recognition algorithm, including a preprocessing algorithm, a feature vector extraction algorithm, and a clustering algorithm, is designed and integrated to serve the system for further automatic controllability. By utilizing this hybrid actuator mode, the robotic exoskeleton is able to train each finger joint independently in a passive training mode and maintain the damping force output within acceptable ranges for different levels of muscle strength. Importantly, with further optimization and upgrades, we deduce that this system has excellent potential applications for finger rehabilitation.

Keywords

• Rehabilitation
• signal processing
• pattern recognition
• stroke
• robotics