Bayesian Terrain-Based Underwater Navigation Using an Improved State-Space Model

Abstract

This paper focuses on terrain aided underwater navigation as a means of aiding an inertial navigation system. It is assumed that a prior map is present and Bayesian methods are used to estimate the position of the vehicle. Traditionally this has been done using a crude low-dimensional model in the Bayesian filters. An improved state-space model is introduced, implemented in a particle filter/sequential Monte Carlo filter and tested on real AUV (autonomous underwater vehicle) data. Compared to conventional filter models, the new model yields smoother, slightly more accurate results, though problems with overconfidence occur.