Research On System Design And Interaction Control Strategy For Lower Limb Exoskeleton

With the rapid development of China's economic strength and technological level,wearable lower extremity has been widely used in the fields of helping the elderly and the disabled,medical rehabilitation,etc.It can significantly improve the exercise capacity and quality of life of people with reduced limb function.In recent years,the research on lowerextremity exoskeleton has achieved certain results,but there are still some problems to be solved in the aspects of system structure design,gait recognition,human-machine interaction and collaborative control.This article focuses on lowerextremity exoskeleton system,from the design and integration of lowerextremity exoskeleton system,human gait recognition and intention recognition,control strategy and performance evaluation of exoskeleton system.This article first proposes the design indexes of the lower extremity exoskeleton system based on the principle of human lower extremity movement.The lower extremity exoskeleton HEXO(HIT EXOskelton)is developed based on design index.Anthropomorphic and non-anthropomorphic joints are designed at the joints,which can solve the problem of knee singularity in the initial state of exoskeleton startup.In addition,a series elastic body based on double parallel springs has been designed and can be used as a series elastic actuator(SEA)to improve joint flexibility.The sensor network of the lower extremity exoskeleton system uses dual-channel CAN communication,which can collect multi-source sensor information and calibrate some specific sensors.Through modeling exoskeleton SEA joints,the experiments with different parameters of multifrequency high and low impedance were carried out.Experimental results show that SEA joints have higher working bandwidth.Higher tracking accuracy can be obtained when using high impedance parameters,and human-machine interaction is more comfortable when using low impedance parameters.In terms of gait recognition,this article divides it into gait pattern recognition and gait phase division.Gait pattern recognition algorithm is built based on long short-term memory model and convolutional neural network.The proposed algorithm is used to compare with common recognition algorithms in several typical gait patterns(squat down,stand up,walking on the ground,up stairs,down stairs).The results show that the proposed gait pattern recognition algorithm works best,and the average accuracy of its recognition can reach 97.78%.Based on fuzzy logic rules,the gait phase is divided into four phases(initial contact,middle stand,initial swing,swing).The effectiveness of the algorithm was verified through experiments.Through the six-dimensional force collected at the human-machine interaction,a mapping model of human-machine interaction force and joint trajectory.The Kalman filter is used to predict the interaction force signal,which can be realized human intention identification.The lower extremity exoskeleton dynamic model was established and decoupled.According to the movement characteristics of the exoskeleton in different phases,a dualmode hybrid control strategy is proposed,which can accurately follow the trajectory of the human body's movement intention and realize human-machine integrated coordinated movement.In the stand phase,the exoskeleton joint needs to bear a large load,so the adaptive impedance control strategy is used to improve the stability and anti-impact ability of the exoskeleton system.In the swing phase,the exoskeleton joint has a larger range of motion,so the active disturbance rejection control and fast terminal sliding mode control improves the response speed and tracking accuracy of the exoskeleton system.In addition,a smoothing strategy is proposed during the transition between the stand phase and the swing phase.Through control simulation and active and passive tracking experiments,the algorithm proposed in this paper is compared with PID algorithm and ADRC algorithm,which shows the superiority of the control algorithm proposed in this paper.In order to evaluate the subjective wearing feeling of HEXO of lower limb exoskeleton and the effect of objective assistance,an evaluation method based on subjective and objective evaluation indexes were established.Subjective evaluation indexes include the convenience of putting on and taking off,the comfort of wearing,the flexibility of movement,and the power of support.Objective evaluation indexes include tracking error,human-machine interaction and heart rate changes.Through the experiments of four experimenters standing up on the squat,walking on the ground and going up and down the stairs,the subjective and objective indexes were evaluated.The subjective evaluation mean score is 86.8,the root mean square error of the hip and knee motion is within 2°,and the average power-assisting effect in the Z-axis can reach nearly70%.Compared with the anthropomorphic HEXO,the non-anthropomorphic HEXO can increase the effect of power assistance by nearly 5%.According to the comprehensive evaluation method,HEXO's "interaction performance" and "assistance performance" were evaluated,and their average scores were 88.9 and 84.4,respectively.The experimenter wearing exoskeleton walks on a treadmill and collects heart rate changes under four states(without HEXO,with HEXO non-driving,with HEXO nonanthropomorphic,with HEXO anthropomorphic).The results show that wearing the driving HEXO state can save the human energy consumption,and the nonanthropomorphic state HEXO has a better assisting effect.