Dynamic parameter identification and control of legged robots'single-leg model considering joint friction characteristics
XIAO Haoyang
QIAN Letian
LUO Xin
Abstract：[Objective]The increasing of the motion speed and the manoeuvrability of legged robots poses high level requirements for joint tracking accuracy.The joints of legged robots often work in the conditions of low speed,small stroke and frequent direction switching.The presence of transmission friction has a strong nonlinear influence on the joint control.Accurately identifying the nonlinear friction of its joints is very important for improving the joint tracking accuracy of legged robots.[Methods]For the low-speed motion conditions of joints,the motion data of a single leg of the heavy-load quadruped robot was recorded by the driver,and the data was processed on a personal computer.A segmented fitting model of joint friction was built based on the measured friction characteristics of joints under low-speed motion conditions,and the joint friction and dynamic parameters of single leg of legged robots were identified.Based on the identification results,a feedforward-feedback control strategy was carried out.[Results]Tests on a heavy-duty quadruped robot indicate that the prediction error of joint torques utilizing the proposed method is reduced by about 40%,and the trajectory tracking accuracy is improved by more than 60%,comparing to those obtained by the existing identification and control method based on the linear friction model.
Keywords：Single leg control of legged robotsJoint friction fittingParameter identification of the dynamic modelFeedforward-feedback controller design
Publication Date：2025-06-15
Online Publishing Date：2026-09-12(First online date of this platform, not the publication date of the document)
Pages：9( 10-18 )