Research On Seven Degrees Of Freedom Upper Limb Exoskeleton Robot System With Cable-Conduit Transmission

The motor ability of the elderly and the physically injured patients are relatively weak,and they can be named as weak persons.As for weak persons,the rehabilitation can effectively enhance their motor ability and is helpful for the recovery of body function,thus improving their qualities of life.As most of the human motions in daily life are related to the upper limb,then the recovery and assistance of body function of upper limb are vital to the weak persons.The upper limb exoskeleton robot can be used to achieve the rehabilitation of upper limb,motion assistance and function expansion of weak persons.However,some problems such as the heavy weight of robot,the weak comfort,the easy shifting,and the difficult in recognizing motion intention still exist.Therefore,an upper limb exoskeleton robot was designed in this work.In addition,the signal prediction algorithm,the motion intention estimation algorithms,and the proportional myoelectric control algorithm were proposed to solve these problems.The 7-DOFs upper limb exoskeleton robot(CABXLexo-7)with the cable-conduit transmission mechanism designed in this work was a hybrid mechanism that was compact.The shoulder joint part and the elbow joint part of the exoskeleton were designed based on the epicyclic gear trains structure.The other parts were the traditional serial structure,then the exoskeleton was a hybrid mechanism.The contact surface between human and machine(the upper limb and the forearm limb)was a torus to increase the area and the whole mass distribution was more even by combing the hybrid mechanism,thus solving the problem of easy shifting and enhancing the comfort and safety.To avoid mounting the DC motors,the drives and the controller on the main body of the exoskeleton,the cable-conduit transmission was applied to separate the driving units from the exoskeleton,and the whole weight of exoskeleton on the body was 3.2kg which was lighter than many existing upper limb exoskeletons.The research on the human motion prediction which can be applied to predict the tendency of human motion is important to the control of exoskeleton.The adaptive multiple oscillators linear combiner(AMOLC)algorithm based on the nonlinear Hopf oscillator and the theory of central pattern generator(CPG)was proposed to predict the joint angle.The AMOLC algorithm was inspired by the frequency division idea and it was composed of several nonlinear Hopf oscillators linearly by distributing the adaptive weights to these oscillators.To increase the accuracy of prediction,the reference formula for the value selection of vital parameters was deduced in this work.The experimental results showed that the AMOLC algorithm could be applied to predict the motion with high accuracy,well robustness,and fast execution time.This work mainly explores how to apply the sEMG well to estimate the human intention motion accurately and quickly.The idea of weighted features and time-delayed features were proposed.To estimate human motion from sEMG,the grey features weighted support vector machine(GFWSVM)algorithm based on the theory of grey correlation degree and support vector machine was proposed.To improve the computational speed when the multiple time-delayed features of sEMG were applied,the random forests(RF)using multiple time-delayed features(MTDF)algorithm proposed in this work was based on the random forests.The experimental results verified that the accuracy of motion estimation was high and the computational speed was fast by using these two algorithms.The joint angle tracking control method based on the PI controller and the compensator of hysteresis was proposed to compensate the hysteresis of the cableconduit transmission,thus avoiding to building the complex dynamic model of the seven DOFs upper limb exoskeleton robot.Furthermore,the proportional myoelectric control algorithm based on the time-delayed integrated signal of sEMG was proposed.The feature extraction of sEMG by using the analog circuit was proposed by this work to increase the speed of calculation,and the idea of time-delayed features was also combined.Then we can balance the accuracy of motion intention recognition and the calculation speed well.The method was based on the PID method,therefore,the computation complexity of it was small and the motion tracking performance and the robustness were well.Three experiments were designed in this work to evaluate the performance of passive rehabilitation,motion assistance,and the assistive motion assistance of upper limb exoskeleton robot.The experimental results showed that the designed upper limb exoskeleton combined with the AMOLC algorithm,the GFWSVM algorithm,the RF using MTDF algorithm,and the proportional myoelectric control algorithm can be applied to achieve the motion intention recognition and the motion assistance with high accuracy and fast execution time.