A new version of linear Kalman filter for GPS navigation is reported in the recent literature. This new version of recursive linear Kalman does not need any linearisation for non-linear GPS measurements and enables accurate estimation of the user position. In this paper, robust version of LKF is presented for GPS navigation with outlying errors in the GPS measurements due to failed satellites or unmodelled errors. In this approach, the innovation of the new LKF is modified using the robust statistics concept to obtain a more accurate state estimate in the presence of outlying errors. Performance of the proposed robust LKF in comparison with the conventional new LKF is demonstrated through simulation results.
Keywords: GPS, linear Kalman filter, robust statistics