Research On Semg-Based Anthropomorphic Variable Impedance Control Of Manipulator

In the unknown environment,the manipulator can achieve precise impedance control during complex contact tasks,but it is difficult to dynamically adjust its impedance characteristics to optimize performance.Humans have good upper limb impedance dynamic adjustment skills to complete complex contact tasks.This paper combines the advantages of the manipulator and the upper limbs of the human body to perform complex contact tasks,by using surface electromyography(sEMG)to identify and classify the endpoint stiffness of the human upper limbs in real time,through the transmission of human-machine skills,the dynamic impedance parameter adjustment of the manipulator is realized by the anthropomorphic variable impedance control strategy.First,a method for calculating the joint stiffness of the upper limb of the human body is studied.A theoretical model for calculating the upper limb stiffness of the human body is established,and the projection gradient descent method is used to solve the upper limb endpoint stiffness.Furthermore,a simplified model of 7-DOF of human upper limbs is established.The spatial position of the upper limb skeletal points is obtained in real time by Kinect.The real-time joint angle of the upper limbs of the human body is solved based on the space vector method and the inverse kinematics of the robot.The endpoint stiffness is converted to joint stiffness by a Jacobian matrix.Secondly,the sEMG-based endpoint stiffness of the human upper limb measurement system is established,and a weighted voting prediction model is proposed to realize the upper limb end stiffness classification.Based on the mechanical disturbance principle and the real-time measurement method of human upper limb configuration,a set of human upper limb end stiffness measurement system including KUKA robot arm,8 sEMG electrodes,six-dimensional force sensor and Kinect was established to realize the measurement of the endpoint stiffness of the human body.The k-means++ algorithm is used to cluster the end stiffness of human upper limbs.Based on the accuracy rate and confusion matrix as the evaluation index,the classification performances of human upper limb sEMG by k-nearest neighbor algorithm,support vector machine,random forest,gradient descent tree and XGBoost are analyzed,and then the weighted voting prediction model is proposed to realize the classification of the endpoint stiffness of the human upper limb.Thirdly,the sEMG-based anthropomorphic variable impedance control strategy is proposed.By using the EMG signal to classify the endpoint stiffness of the human upper limb and combining the Kinect to measure the human body's upper limb Jacobian matrix,the prediction of the joint stiffness of the human body is realized,thereby the manipulator variable impedance control of the manipulator is realized.The simulation model is established by using MATLAB and ADAMS,the performance of the joint fixed impedance control strategy and the sEMG-based anthropomorphic variable impedance control strategy are analyzed,and the effectiveness of the sEMG-based anthropomorphic variable impedance control strategy is verified.Finally,an experimental platform based on robot astronauts was built.Based on the joint fixed impedance control strategy and the sEMG-based anthropomorphic variable impedance control strategy,the anti-interference experiment of the mechanical arm is carried out to verify the effectiveness of the sEMG-based anthropomorphic variable impedance control strategy.