Iterative identification of robot dynamic parameters using grassland grazing algorithm
The accurate dynamic model parameters are essential for robot admittance control,trajectory tracking,and force control.However,the commonly used least square method fails to meet the high precision requirements of dynamic parameter identification.In this paper,we propose using the grassland grazing algorithm to iteratively update the weights of each sampling point in order to avoid outliers in the observation data.Additionally,we utilize the fmincon function to optimize the excitation trajectory used in the parameter identification experiment,aiming to fully stimulate the dynamic characteristics of each connecting rod of the robot.A control experiment is designed to explore how optimizing the excitation trajectory influences parameter identification.Experimental results on a 6-DOF robot validate our identification method:compared with weighted least square method,our proposed method reduces average residual error by 18.12％when predicting driving torque for each joint of the robot.This also provides a new approach for torque tracking control.
万方数据
维普
