Hybrid Rehabilitation System with Motion Estimation Based on EMG Signals

Patients with upper limb paralysis undergo various types of rehabilitation to reconstruct upper limb functions necessary for their return to daily life and social activities. Therefore, it is necessary to develop an effective rehabilitation support system using robotic technologies. In this study, we propose an EMG-driven hybrid rehabilitation system based on the estimation of intended motion using a probabilistic neural network. In the proposed system, the developed robot and functional electrical stimulation are controlled by estimating the patient's intention, which enables the intuitive learning of the appropriate control abilities of joint motions and muscle contraction patterns. In the experiments, hybrid and visual feedback training were conducted for pointing movements of the wrist joint of the non-dominant hand. The results confirmed that the proposed method provides effective training and has great potential for use in rehabilitation.