Adaptive choice of scaling parameter in derivative-free local filters

The paper deals with adaptive choice of the scaling parameter in derivative-free local filters. In the last decade several novel local derivative-free filtering methods have been proposed. These methods exploiting Stirling's interpolation and the unscented transformation are, however, conditioned by specification of a scaling parameter significantly influencing the quality of the state estimate. Surprisingly, almost no attention has been devoted to a suitable choice of the parameter. In fact, only a few basic recommendations have been provided, which are rather general and do not respect the particular system description. The choice of the parameter thus remains mainly on a user. The goal of the paper is to provide a technique for adaptive choice of the scaling parameter of the derivative-free local filters.