GPS navigation with increased immunity to modeling errors

Abstract

A robust extended Kalman filter (EKF) is presented for GPS navigation with outlying errors in the GPS measurements due to failed satellites or unmodeled errors. In this approach, the innovation step of the conventional EKF 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 EKF in comparison with the conventional EKF, and adaptive EKF is demonstrated through simulation results.