Application of a bounded‐error on‐line estimation algorithm to robotics systems

Abstract

Abstract This paper shows how bounded‐error on‐line estimation algorithms can be applied in the context of parameter identification of mechanical structures. the modified exponentially weighted recursive least‐squares (MEW‐RLS) estimation algorithm proposed by Canudas de Wit and Carrillo is used in connection with models of mechanical systems with state‐dependent disturbances. These disturbances are provoked on one hand by simplified friction models and on the other hand by the particular model formulation of the multivariable Lagrange mechanical systems. A priori knowledge on the physics of these systems is used to derive upper bounds on the model disturbances, which are needed for the application of the algorithm.