IDENTIFICATION OF ROBOT DYNAMICS: AN APPLICATION OF RECURSIVE ESTIMATION.

Abstract

To synthesize robust robot parameter identification algorithms, we outline the fundamental properties of the Newton-Euler (N-E) and Lagrange-Euler (L-E) formulations of robot dynamics. We transform the nonlinear (in dynamic parameters) N-E dynamic robot model into the equivalent linear (in dynamic parameters) L-E dynamic robot model. We cast the L-E torque/force error model into the series and parallel identifier structures for on-line and off-line robot parameter estimation. To illustrate our approach, we identify (in simulation) the dynamic parameters of the cylindrical prototype robot and the three degree-of-freedom positioning system of the Stanford manipulator. Our identification algorithm is directly amenable to the real-time identification of the pay-load inertial characteristics and the dynamic frictional coefficients for precise trajectory control.