Research On Rehabilitation Robot Control Based On SEMG

Motor function reconstruction and rehabilitation after stroke causes widespread concern of researchers in the scientific community. Rehabilitation cycle of patients after stroke is long, and rehabilitation proceeding needs rehabilitation physicians help patients to do exercises manually. So, it's significant for stroke patients combining robotics technology with rehabilitation technology. At the present stage, most of rehabilitation robots have three types of rehabilitation models, i.e. passive model, active model and impedance model, among of that, active model is the best method because initial action will improve motor function of patients more quickly in the reconstruction and rehabilitation of motor functionSurface electromyography (sEMG), a kind of biological signals generated with muscle contraction, can reflect the neuromuscular activity by which patient's intention can be estimated. Meanwhile patients can carry out rehabilitation exercise according to their intention.In the dissertation, a upper limb rehabilitative robot, which is suitable for unilateral limb motor function damaged hemiplegia patients, is proposed. Robot system drives paralysed side of patients according to their own intention estimated with sEMG of healthy side. The main contents are as follows:Firstly, this dissertation elaborates generation mechanism and characteristics of EMG then describes the surface EMG and joint angle acquisition system, which can collect four channels surface EMG Also, we extracts five kinds of electromyography features on time-domain.Secondly, this dissertation studies on the method of EMG-based joint quantitative estimation, by which a quantitative model, accompanying with extracted features of sEMG as input and joint angle as the output, based on support vector machine and BP neural network respectively is proposed.Thirdly, this dissertation establishes a rehabilitation robot system to verify the contents above. According to the D-H method, to the dissertation has established kinematics model of the rehabilitation manipulator and established the dynamic model based on Lagrange equations. Then, dissertation has described the basic principles of the impedance control, and two control methods, including force-based impedance control and position-based impedance control. At last, the impedance parameters are estimated by experiments.Finally, this dissertation establishes the integrated system of upper rehabilitation robot utilizing patients' active control method which estimates intention of patients online based on sEMG, to verify the feasibility of rehabilitation system.