On-line robot dynamic identification based on power model, modulating functions and causal Jacobi estimator

Abstract

This paper estimates robot dynamic parameters by means of power model associated with modulating functions, which avoids measuring or calculating the joint acceleration. At the same time, an advanced causal Jacobi derivative estimator is applied in order to get on-line robust derivatives from noisy measurements. In the end simulation results on two degrees of freedom planar robot are presented and comparisons with traditional off-line identification method are drawn.