Research On Intelligent Interaction-rehabilitation-assessment Strategy Of End-traction Upper Limb Rehabilitation Robot

Objective: Stroke-induced death and disability have become the most burdensome diseases in the world.Most stroke patients have upper limb motor dysfunction,which greatly affects their daily life and psychological burden.Traditional upper limb rehabilitation training is mainly manual,which is time-consuming and labor-intensive.This paper addresses the problems of inefficient rehabilitation strategy,poor interaction effect,insufficient safety and lack of intelligent rehabilitation assessment of current upper limb rehabilitation devices,and conducts research on intelligent interactionrehabilitation-assessment strategy of end-traction upper limb rehabilitation robot for stroke patients.Methods: 1.To address the problem of variability in upper limb training trajectory and training intensity among different patients,an end-traction upper limb rehabilitation robot direct teaching technique is proposed for passive rehabilitation.The algorithms of gravity compensation and constraint step conversion in the robot operating system(ROS)enable the end of the robot arm to comply with the drag force of the rehabilitation therapist to track and record the teaching trajectory at the same time.Moveit is used to write the trainer reproduction node,which is used to adjust the number of training sessions,training speed and rest time for rehabilitation training and then perform targeted training.The kinematic solution and trajectory planning of the trajectory are carried out to achieve the effect of personalized training.2.On the basis of the trajectory,an adaptive force field active rehabilitation strategy based on conductance control is proposed to further improve the flexibility,safety and human-machine interaction of the rehabilitation robot.The interaction suppleness is increased by introducing the conductance control.For the dragging trajectory,the best-fitting equation is determined by using recursion,and then the trajectory force field is realized by adjusting the desired position in real time through the end position;for the regular trajectory force field is realized by using a simpler planar segmentation method.The force field size is adaptively adjusted by means of positionforce control,and force compensation is performed in the tangential direction of the desired position to achieve active rehabilitation training with adaptive force fields.Based on this,quantitative rehabilitation indexes such as training range,training smoothness,training accuracy and training participation are proposed,and the mathematical expression of rehabilitation assessment is obtained by regression to achieve quantitative rehabilitation assessment of upper limb motor function while conducting active rehabilitation training.3.Based on the above control strategy,an interactive upper limb rehabilitation robot imitating therapist traction based on motion synergy analysis is proposed.rehabilitation strategy.By building a recurrent neural network and a muscle mathematical model,the abstract features in the muscle synergistic motion information are extracted from the surface EMG signals for force/stiffness regression at the end of the upper limb,combined with the force field constrained interactive control strategy,and then the anthropomorphic rehabilitation task of the upper limb rehabilitation robot is realized.Results: 1.5 healthy subjects completed the demonstration experiments in different spaces respectively,and the demonstration trajectories were smooth and took a short time.Subsequently,variable intensity training was successfully performed.2.The adaptive force field had higher flexibility and safety than the constant force field,and the booster and resistance training was achieved by setting the compensation force;8 subjects participated in the quantitative rehabilitation assessment experiment,and the best training effect was obtained when K=250N/m,and the mathematical expression of the rehabilitation assessment was obtained by fitting the index.3.6 subjects performed the upper limb continuous force/stiffness estimation The performance of synergy-based force/stiffness estimation was R2=0.981 and R2=0.851,respectively.subjects successfully performed myoelectric active training under force-free sensors and rehabilitation training under two abnormal training conditions.Conclusions: 1.The direct teaching system can help trainers to teach multi-degree-offreedom upper limb training trajectories for different subjects in a simple,fast and stable manner,and to conduct rehabilitation training of different intensities by reproducing the trajectories,which is simple to operate,targeted and accurate.2.The adaptive force field added to the planar arbitrary training trajectory can improve training flexibility,safety and training targeting,and realize multiple rehabilitation modes.The proposed rehabilitation assessment indexes can better characterize the upper limb motor function and achieve real-time quantitative assessment.3.The continuous force/stiffness estimation performance of the upper limb reaches the state-of-the-art level;the combination of synergistic neural interface technology and flexibility control largely improves the effect of human-robot interaction training,which is of reference significance for anthropomorphic rehabilitation tasks of the robot.