Three-Degree-of-Freedom Upper Limb Rehabilitation Training System Based On Adaptive Assist-as-Needed

Stroke is one of the major diseases that endanger the health of the elderly.Studies have shown that rehabilitation training can promote the recovery of motor function in stroke patients.Traditional rehabilitation training requires one-on-one training between therapists and patients,but the number of therapists is insufficient and the treatment cost is expensive.The emergence of rehabilitation robots solves the above problems.At present,the research on rehabilitation robots and their control algorithms has become a research hotspot at home and abroad.In this paper,a three-degree-of-freedom upper limb rehabilitation training system based on adaptive assist-as-needed control is designed.Firstly,the mechanical structure of the threedegree-of-freedom upper limb rehabilitation robot is designed,which is mainly composed of the traction manipulator,the support manipulator and the wrist structure.Then,the forward and inverse kinematics analysis,statics analysis and motion range analysis are carried out on the mechanical structure of the rehabilitation robot,and the hardware measurement and control system of the rehabilitation robot is designed.An adaptive assist-as-needed control algorithm is designed.Firstly,a method for evaluating the movement status(Methods of Movement Status,MMS)is designed,which comprehensively measures the patient’s movement status according to multiple parameters,and then control system adjusts the robot’s auxiliary force according to the patient’s movement status,so as to realize the adaptive assist-as-needed control of the position.The adaptive assist-asneeded control algorithm consists of a current PI control module,a force-current conversion module,an adaptive assist-as-needed control module,and a position impedance control module.The interactive software is designed based on MFC,including functions such as data analysis and display,algorithm parameter adjustment,UDP communication and inter-process communication.The virtual scenes are designed for passive training mode,active training mode and power-assisted training mode respectively,and the realization process and data transmission method of the virtual scene are analyzed.Finally,the verification experiment of the rehabilitation robot system is carried out.Experiment 1 fitted the relationship between torque and current through experiments.In experiment 2,by measuring the average error of the robot in two different trajectories,the trajectory tracking error is calculated to be 17.6mm.In experiment 3,by comparing with the traditional algorithm,it was calculated that the auxiliary force of the algorithm in this paper was reduced by 61.4% and compensated for 85.3% of the interfering external force,thus improving the patient’s initiative and system compliance.Experiment 4 verifies the effectiveness of the three modes by calculating the position error and mean square error of the three modes under the sinusoidal trajectory,and the results show that the three modes are all effective.Experiment 5 verifies the tracking accuracy of the output force of the wrist,and the tracking error of the output force is calculated to be 0.58 N.Therefore,the system has the characteristics of high trajectory tracking accuracy,can improve the patient’s initiative and system flexibility,multiple training modes,and high wrist output force tracking accuracy,which meets the requirements of rehabilitation.