A Novel Multi-Hypothesis Filter for Terrain Aided Navigation

Abstract

The susceptibility of GNSS signals has led to the development of advanced aiding systems in which information obtained from external sensors is fused with the inertial navigation solution to maintain navigation continuity in GNSS-denied environments. In terrain-aided navigation, the aim is to solve the horizontal position ambiguity of the platform with the help of pre-installed digital terrain elevation maps and estimated terrain height information obtained by the platform's barometer and distance measuring sensor readings. In this paper, we propose a novel multi-hypothesis terrain-aided navigation framework where horizontal position measurements are generated along with their covariance matrices from the estimated terrain heights. These measurements are fused with appropriate navigation hypotheses in a closed-loop configuration where augmented navigation error state estimates are fed into the navigation solution of the corresponding hypothesis. The number of hypotheses and their covariances change adaptively based on the terrain under the flight path preventing deviation from the true position for multi-modal or non-informative regions. The simulation results show that, unlike conventional TAN methods, the proposed method is able to yield accurate positioning estimations even when low-end IMUs are used. Furthermore, the proposed method is found to be effective in the presence of substantial initial errors in position, velocity, and heading.