Research on Accurate Identification of Robot Gravity and Friction Forces
This paper presents a robot gravity and friction force identification method based on current esti-mation torque,which has the advantages of simple,fast,and accurate identification.A linear regression model on the robot's gravity term parameters is established,and the identifiable gravity term parameters are extracted based on the numerical method of QR decomposition.A linear interpolation excitation trajectory with periodic symmetric splicing parabolic transitions is designed,and the gravity term parameters are esti-mated using the least squares method.The friction forces of each joint of the robot are modeled and identi-fied based on the Coulomb-viscous model and Stribeck model,which improves the fitting accuracy of the fric-tion model.Experimental results show that when the robot operates at low speed,the theoretical torques calcu-lated based on the identified parameters can fit the actual driving torques of the robot well,with an average absolute error and root mean square error within 2 N·m,verifying the accuracy of the identification results.
robotgravity identificationfriction force identificationtrajectory planning
NETL
NSTL
万方数据
