Grid Design for Efficient and Accurate Point Mass Filter-Based Terrain Referenced Navigation

Abstract

This paper proposes an optimal grid design method for point mass filter-based terrain referenced navigation (PTRN) by considering the process and measurement uncertainties or noise to efficiently obtain accurate results. The estimation performance of the point mass filter (PMF) is generally known to improve when very large support and a high-resolution grid are implemented. However, when this condition is applied to the PTRN, the algorithm cannot be executed in real-time due to the high computational load. In addition, even though the grid condition is improved, the filter accuracy is limited by the given process and measurement noises. Therefore, it is possible to perform efficient and accurate PTRN by finding the minimum number of grid points that can achieve the maximum performance. In this paper, a grid design method is carried out by considering each noise, and the selection logic between the two design results are proposed. By applying the proposed grid design method, it is possible to obtain almost the same accuracy as the results that are obtained when a very high resolution is applied, with a much lower computational load, and it is expected for the highly accurate PTRN to be available in real-time.