A Motor Recovery Training and Evaluation Method for the Upper Limb Rehabilitation Robotic System

Stroke can cause acute damage to blood vessels in the brain, which often leads to hemiplegia and imbalances in mobility. It is a challenge to develop a motor recovery training and evaluation method with less help of therapists. In this article, a motor recovery training and evaluation method for the upper limb rehabilitation robotic system is proposed. This system has two rehabilitation units, one is active, and the other is passive. The developed rehabilitation robotic system includes an exoskeleton rehabilitation robot and PHANTOM1.5. The patients do rehabilitation training with the help of a robot in the passive unit, and a virtual reality game is designed in the active unit. Patients with mild motor impairments observe the virtual reality game interface while manipulating the PHANTOM to do rehabilitation training. Three experiments are proposed in this paper. The fuzzy neural network (FNN), spring-damper model, and the method to evaluate the training trajectory are designed and validated. The surface electromyography (sEMG) signals and grip force during rehabilitation training are collected to set up an FNN and achieve evaluation. The accuracy of the network is 0.96 which is calculated in validation set. In Section IV , the rehabilitation evaluation method is compared with the state of art on rehabilitation evaluation method. The method proposed in this article can reach high accuracy. It is easy to use and understand for patients even without the help of therapists. The problem of lacking therapists can be solved to some extent by the proposed upper limb rehabilitation system.