Research On The Intelligent Adaptive Impedance Control Of Human Body-Oriented Robot

When robot performs massage,medical rehabilitation,and other tasks on human body,the uncertainty of human body's mechanical properties may make it difficult to control the force accurately in impedance control.This paper proposes a self-tuning impedance control system that incorporates Particle Swarm Optimization parameter identification algorithms.Through the identification of human body dynamics,the self-tuning controller will guide the feed forward prediction,and get the feedback-adjustment of the reference position to achieve accurate force and position control in an unknown human body environment.At the same time,in view of the problem that the Particle Swarm Optimization parameters coupling affect the operation efficiency of the identification algorithm,a Graph Knowledge Transfer Learning algorithm for Particle Swarm Operation parameters is proposed.The historical information is used to obtain empirical information,and it is learned and applied to the target task,which will speed up the convergence speed of the identification algorithm.The research contents mainly include:(1)Through analysis of human body's stress-deformation data,a non-linear human body dynamics model is established to describe the human body's unique biomechanical characteristics,and the parameters of the model is identified based on Particle Swarm Optimization algorithm.The Particle Swarm Optimization algorithm will overcome the difficulties of coupling and non-linear feature of human body dynamic model parameters;(2)According to the optimization requirements of Particle Swarm Operation algorithm parameters,based on the idea of transfer learning,the source task screening algorithm and the transfer function learning algorithm is proposed to learn experience information from historical job and apply them to the optimization tasks.This will reduce the iterations times of identification algorithm and improves the identification efficiency of the Particle Swarm Optimization algorithm;(3)An adaptive impedance control system is proposed based on the nonlinear human body dynamics model and Particle Swarm Optimization identification algorithm,the feedforward prediction of the reference position in impedance control is performed,and the reference position is adjusted based on the force-feedback and self-tuning control,which will guarantee the force and position control effect while achieving compliant contact in unknown human body environment.The experimental of the research content was performed on human body.The average fitting error of human body dynamic model in dynamic model fitting experiment was less than 0.2N;In the adaptive impedance control experiment,the robot can move along the planned working path while keeping the average absolute error of the force below 0.2N and the variance of the contact force below 0.4N~2,achieving precise force and position control effect;In the transfer learning optimization experiment of Particle Swarm Operation parameters,under the same identification accuracy requirements,the Particle Swarm Operation identification algorithm after optimization reduces the average iterations times by51.82%,and the identification efficiency is significantly improved.