3D map matching using a few rangefinders and an uncertainty model considering the arrangement of buildings for urban canyon navigation

Abstract

In this work, a complementary navigation algorithm for a Global Navigation Satellite System (GNSS) in urban canyons is presented. The proposed system uses a few unidirectional rangefinders to determine positions by 3D map matching. For 3D map matching problem, the observation model can be defined as a discontinuity piecewise linear function. Owing to discontinuities, a typical linearization-based tightly coupled integration algorithm is inadequate. Therefore, this paper presents an overall algorithm flow that loosely couples the position determined by 3D map matching with inertial navigation and assesses the effectiveness of this method. The small number of measurements implies that we need a matching method different from the used by the conventional LADAR-based system, where thousands of measurements are possible. In other words, conventional matching techniques such as the widely used ICP (iterative closest point) approach for LADAR are inappropriate. Therefore, we propose a matching method based on the MSD (mean squared difference) index suitable for the proposed system. An uncertainty model of the matching results is also proposed. Given that the accuracy of the matched position is mainly influenced by the arrangement of the surrounding buildings, the proposed model is designed to reflect the geometric properties of the buildings. As a first proof of concept, we conducted a Monte-Carlo simulation for a comparison of different measurement update methods and a comparison of uncertainty models. The simulation results show that the proposed algorithm outperforms the conventional algorithm.