Identification of dynamic parameters of an industrial robot using a recursively-optimized trajectory

Abstract

Dynamic parameters of a robot affect the performance of advanced control schemes significantly. In this study experiments to identify the dynamic parameters of AT2 robot are carried out. The excitation trajectory for identification is parameterized with Fourier series. The trajectory is optimized to minimize effects of uncertainty using condition number as the index. Recursive optimization is proposed so that the trajectory is robust from selection of the initial values and better optimization result can be obtained. The dynamic parameters of AT2 robot is estimated using the optimized excitation trajectory and weighted least-square method.