Probabilistic Map Matching for Robust Inertial Navigation Aiding

Abstract Robust aiding of inertial navigation systems in GNSS-denied environments is critical for the removal accumulated error caused by drift and bias inherent sensors. One way to perform such an uses matching geophysical measurements, as gravimetry, gravity gradiometry or magnetometry, with a known geo-referenced map. Although simple concept, this map-matching procedure challenging: The measurements themselves are noisy, their associated spatial location uncertain, may match multiple points within map (i.e., non-unique solution). In paper, we propose probabilistic multiple-hypotheses tracker solve problem allow robust aiding. Our approach addresses both locally, via data association, temporally incorporating underlying platform kinematic constraints into tracker. estimated position from output then integrated state using unscented Kalman filter. Additionally, present statistical measure local information density — feature variability use it weight covariance proposed algorithm. effectiveness robustness algorithm demonstrated scenario involving gravitational matching.