Adaptively tuning the scaling parameter of the unscented Kalman filter

Abstract

This paper describes a new approach to adaptively tuning the scaling parameter of the unscented Kalman filter. The proposed algorithm is based on the idea of moment matching and is computationally inexpensive, allowing it to be executed online. Two nonlinear filtering problems are used to numerically compare the performance of the proposed algorithm with the performances of recently published adaptive unscented Kalman filters. An unscented Kalman filter enhanced with the proposed adaptive algorithm outperformed the other adaptive filters in both numerical problems. © 2015 Springer International Publishing.