Non linear optimum filter based smoothing Interacting Multiple Model for GPS navigation system

Abstract

An Interacting Multiple Model Unscented Two Filter Smoother (IMM-UTFS) approach for GPS navigation system is introduced in this paper. The Unscented Kalman Filter (UKF) propagates its state estimate and covariance through unscented transform without any need of linearization. The Interacting Multiple Model (IMM) algorithm obtains its estimate by combining the individual estimate from a number of parallel filters matched to different motion models of the vehicle. This paper adopts the Unscented Two Filter Smoother to the IMM algorithm to increase the navigation estimation accuracy. The dynamic behavior of the vehicle is analyzed and the simulation results show that IMM-UTFS can improve overall navigation accuracy as compared to traditional filters like UKF and multiple model filters like IMM-UKF.