Parameters Identification of Whole Body Dynamics for Hexapod Robot

Yang Zhang, Yue Gao, Ming Sun

Abstract: Most motion controllers of legged robots are based on dynamics models to ensure stable and fast locomotion on rough terrains. Accurate dynamics parameters can provide a reasonable feedforward signal for the motion controller to improve the response speed of the system. However, the dynamics parameters obtained by computer aided design (CAD) and manual measurement may bring large errors. In this paper, a whole body dynamics parameters identification method for hexapod robots based on least square estimation is proposed. The identification model is established by linear optimization of the leg dynamics equation and the whole body moment model of the Hexapod robot. As the robot moves according to the reference trajectory, the joint torques and ground reaction forces (GRFs) are collected to complete the parameters identification. Experiments in simulation demonstrate the effectiveness of the parameters identification method.