Study On Control Simulation Of Upper Limb Exoskeleton Rehabilitation Robot Based On Mechanomyogram And ROS

There are a huge number of new stroke patients in China each year,and many patients suffer from loss of limb motor function due to stroke,which brings a lot of inconvenience to life.Rehabilitation robots,as an emerging means of rehabilitation treatment,replace rehabilitation therapists to drive patients’ arm movements to achieve passive training,which not only ensures the consistency of repetitive movements,but also reduces the labor intensity of physicians.It is the development direction and research hotspot of rehabilitation engineering.Mechanomyogram Signal(MMG)is the low-frequency vibration signal when muscles contract.As a new biomedical signal,it has broad application prospects in the field of rehabilitation engineering.This subject deeply analyzed the development status of domestic and foreign rehabilitation robots and the characteristics of upper limb rehabilitation after stroke.A 5-DOF exoskeleton rehabilitation robot model for stroke patients’ upper limb rehabilitation training was designed.The coordinate transformation of the kinematic relationship between the joints was obtained by D-H method.The forward kinematics of the model was analysed using Matlab,and the workspace of the robot was calculated.A simulation control environment based on the open sourced robot control system ROS was built,and the control system and the mechanomyogram signal online acquisition system were established.A rehabilitation training model guided by the mechanomyogram signal of the contralateral upper limb is proposed.By collecting the mechanomyogram signal of the contralateral upper limb,the support vector machine algorithm is used to perform shoulder abduction,shoulder extension,internal rotation of the upper arm,and elbow flexion and extension.Five types of forearm rotation/exterior rotation movements are used for online movement pattern recognition,and the patient’s movement intention is obtained to drive the robot model to execute the corresponding movement on the simulation control platform,which realizes the simulation control strategy based on the pattern recognition of mechanomyogram signals.The results of simulation control experiments show that the classifier model can identify the user’s movement intentions relatively accurate,and the average online recognition rate of the five types of upper limb movements reaches 90%.The model moves smoothly in the simulation environment,and can accurately reproduce the pre-planned rehabilitation movement trajectory,laying a simulation foundation for physical construction.