Design Of Soft Rehabilitation Glove-assisted Training System Based On Motor Intention Recognition

Hand motor dysfunction,as one of the common motor dysfunctions after stroke,causes great disturbance to the normal life of patients.Assisted rehabilitation training based on the patient’s motor intention can effectively promote brain motor nerve remodeling and accelerate hand motor function recovery.However,the current training mode of soft rehabilitation gloves is mostly mechanical repetitive training with a single gesture,and it isn’t possible to perform rehabilitation training with multiple gestures based on the patient’s motor intention.To address the above-mentioned problems,this paper designed a soft rehabilitation glove assisted training system based on the mirror image rehabilitation theory.The main research contents and academic results achieved are as follows:(1)A dual-drive multi-joint soft actuator was designed.By analyzing the skeletal structure and motion characteristics of the human hand,the soft actuator dimensions of the five fingers were designed,and the motion space of the soft rehabilitation glove was solved by the D-H matrix method to verify the rationality of dimensional design.On the basis of the dimensional design,a dual-drive multi-joint soft actuator was designed considering the independence and coupling of the motion between the finger joints.The influences of key structural parameters on the bending performance were studied with finite element method,the reasonable structural parameters of the soft actuator were determined.The soft actuator was fabricated by 3D printing mold and casting process,and experiments were conducted to show that the soft actuator could perform two degrees of freedom motion and has well bending performance.(2)A hand motion intention recognition model based on surface electromyographic(s EMG)signals was established.Six hand gestures were selected as the object of motion intention recognition to analyze the correlation between hand motions and forearm muscles,thus the muscles most associated with the six gestures were identified,and then experiments on the acquisition of s EMG signals were carried out.The obtained s EMG signals were filtered and activity segment detected,then features were extracted from the time and frequency domains on different hand gestures.A hand motion intention recognition model based on the combination of time-frequency combined features and support vector machines based on directed acyclic graphs after grid search optimization(GS-DAG-SVM)was established,and the experimental results showed that the model performed well accuracy.(3)The hand rehabilitation assisted training system was built.The system included three parts of functional design and implementation: the upper computer subsystem,the soft rehabilitation glove and the lower computer subsystem.Through testing of the assisted training function,the test results showed that the system could well identify the movement intention of the healthy side hand,and control the soft rehabilitation glove to assist the affected side hand to perform the same kinds of hand movements as the healthy side hand,which realized the purpose of assisting training based on the patient’s movement intention and the effectiveness of this system was demonstrated.