A robust Kalman filter for mars entry navigation

In this paper, a robust Kalman filter is proposed for Mars entry navigation. The multiple uncertainties present in the Mars atmosphere density together with the unmodeled measurement noise, which may cause the divergence of traditional extended Kalman filter, are mainly considered. In order to improve the robustness of traditional Kalman filter, the relationship between estimation error and the Mars atmosphere density uncertainty and unmodeled measurement noise is derived. Besides the trace of posterior variance matrix, the transition matrices of atmosphere density uncertainty and unmodeled measurement noise are considered to establish the performance index. Next, the optimal Kalman gain matrix is determined by minimizing this performance index. Navigation scenario using range measurements from Mars ground beacons demonstrates the improved accuracy of the proposed robust Kalman filter. Furthermore, the contribution of the coefficients which are corresponding to atmosphere density uncertainty and unmodeled measurement noise to the navigation performance are analyzed. The optimal values of these coefficients are also recommended.