An Adaptive Unscented Kalman Filter for Dead Reckoning Systems

The sequential filtering of discrete time nonlinear systems in the presence of unknown noise statistical parameters or time varying noise parameters is studied in this paper. The Sage-Husa statistics estimator is introduced to unscented Kalman filter (UKF), then the online estimation of unknown covariance of noise is completed with recursive operations, a novel adaptive unscented Kalman filter (AUKF) is proposed. The feasibility of this method is proved with a simulating example of dead reckoning (DR) system, and it positioning precision outperforms UKF, extended Kalman filter (EKF).