RBPPFF for robust TAN

Abstract

Terrain-aided navigation (TAN) is a technique for position estimation based on altimeter and digital elevation model measurements. To perform TAN, non-linear filters such as a particle filter (PF) or a point mass filter (PMF) are mainly used owing to the non-linearity of measurements and systems. In particular, owing to the degradation of the global positioning system or sensors and the non-linearity of maps and measurements, an initial state error or sudden state change may occur. The authors propose a Rao-Blackwellised particle-point mass fusion filter (RBPPFF) that performs robustly in the presence of such initial state errors or sudden state changes with low computational complexity in real time. The RBPPFF combines the advantages of a PF and PMF, and it employs Rao-Blackwellisation to efficiently increase the number of states. To develop the RBPPFF, the formulae of the RBPF and Rao-Blackwellised point mass fusion filter (RBPMF) are derived using the same conventions, and a method is presented for an effective fusion of the PF and the PMF. Finally, the performance of the proposed algorithm is demonstrated through Monte Carlo simulation across various environments with filters such as PF, PMF, RBPF, and RBPMF.