Cooperative Unscented Kalman Filler with Bank of Scaling Parameter Values

This paper is devoted to the Bayesian state estimation of the nonlinear stochastic dynamic systems. The stress is laid on Gaussian unscented Kalnian filter (UKF) and, in particular, on a setting of its scaling parameter, which significantly affects the UKF estimation performance. Compared to the standard UKF design, where one scaling parameter per a time instant is selected, the proposed cooperative UKF combines estimates of the set of UKFs each designed with different value of the scaling parameter. The cooperative UKF reformulates the UKF scaling parameter selection task as the multiple model approach, which allows to extract more information from the measurement to provide estimates of better quality as indicated by the numerical simulations.