A Hybrid Robust Information Fusion Scheme for SFNS/XNAV Integration

Abstract

The integration of the spectral frequency-shift navigation system (SFNS) and the X-ray pulsar navigation (XNAV) system provides a promising solution for deep-space navigation, with respect to the complementarity in velocity and position determination. However, since the gross errors involved in the SFNS and the inherent systemic errors in XNAV pertain to different types of measurement uncertainties, it is very difficult to acquire the globally robust state estimation for SFNS/XNAV integration via the existing distributed information fusion methods as usual. This article develops a hybrid robust information fusion scheme with a two-layer architecture to address the aforementioned issue. In the first layer, a Huber M-estimation based extended Kalman filter (HEKF) is designed for the SFNS to restrain the effect of gross errors, while the systemic errors in XNAV are compensated in parallel based on system state augmentation with the augmented extended Kalman filter (AEKF). In the second layer, from the perspective of minimum variance criterion, a multisource optimal data fusion strategy is established to fuse the local robust state estimations output by the HEKF and the AEKF, such that SFNS/XNAV integration is realized in a globally optimal manner. Simulations on a Mars exploration mission have validated that the proposed information fusion scheme exhibits excellent performance in terms of robustness and navigation accuracy in comparison with the existing works.