The Research And Design Of Intelligent Upper Limb Rehabilitation Training System

The incidence of upper limb-related diseases such as shoulders and elbows is increasing.Severe shoulder inflammation,shoulder fractures,golf elbow and other diseases seriously affect people’s health and quality of life.People are increasingly demanding scientific upper limb rehabilitation training.high.Upper limb rehabilitation training is mainly divided into passive rehabilitation and active rehabilitation.Passive rehabilitation refers to the purpose of rehabilitation training for the upper limbs of patients with external traction.Active rehabilitation means that the patient’s own strength against gravity or appropriate resistance to achieve rehabilitation purposes.For common cervical spondylosis,frozen shoulder and other diseases,active rehabilitation can achieve better preventive and therapeutic effects,but active training of patients often has problems such as irregular movements and difficulty in meeting the standard.Therefore,the upper limb rehabilitation training system designed in this paper is divided into two parts: passive and active.For the passive rehabilitation part,this paper designs a complete control algorithm and analyzes and verifies it in Matlab.For the active rehabilitation part,this paper uses OpenPose to perform human body gesture recognition,and studies the reconstruction of 3D human body model by 3D reconstruction technology.Then the dynamic time warping algorithm and random sampling consistency algorithm are used for action flow comparison to evaluate the interactive active training effect.Finally,a software system based on virtual interaction was designed to guide,evaluate and record the active rehabilitation of patients.The simulation results of the five-degree-of-freedom upper limb rehabilitation robotic arm prove that the mechanical arm can meet the requirements of conventional rehabilitation action through kinematics analysis and control.The realization of active rehabilitation software proves that the use of deep neural network-based visual motion capture technology for active rehabilitation training is effective.