A New Federated Filtering Method for Integrated Navigation of Vehicles

Abstract

According to the characteristic of integrated navigation systems of GPS/DR, a new federated filter is proposed, in which the sub-filters are composed by a Kalman filter and an unscented Kalman filter. The information allocation factor is adopted as a form of minimum mean square error (MMSE) base on observation error. By this scheme, not only the output information is fused optimally, but also the computing load and difficulty decrease, and the reliability and error tolerance are improved. The simulation results show that, the position precision is comparative as that of the UKF, and is higher than that of the EKF.