A split-horizon scheme for on-line friction parameter estimation

Friction model identification is in most cases carried out in an off-line manner, where separate measurements are performed to identify the presliding and sliding friction characteristics using specific excitation signals. However, a need exists to update on-line friction parameters to track friction dynamics during normal machine operation. This paper presents some preliminary results of a split horizon estimator where friction parameters are updated on-line during regular point-to-point motions. The split horizon approach alleviates the non-smoothness of the friction model and improves the convergence of the estimation problem. The friction behavior is modeled using the Generalized Maxwell-Slip (GMS) dynamical friction model, which is known to describe all essential friction characteristics in presliding and sliding motions.