Navigation of Unmanned Surface Vehicles Using Underwater Geophysical Sensing

Abstract

Underwater geophysical properties can provide useful information for surface navigation, particularly in situations where a global navigation satellite system is unavailable. Unmanned surface vehicles (USVs) equipped with geophysical sensors can measure certain types of underwater properties related to Earth geophysics. For example, multibeam echosounders can obtain an array of sonar ranges for underwater terrains, and magnetometers can measure geomagnetic vector fields. These measurements can be used to track vehicle poses if pre-surveyed geophysical maps are provided. This paper proposes geophysical navigation of USVs using a multibeam sonar and magnetometer. The navigation algorithm is implemented within a particle filter framework, and we designed observation models for each geophysical sensor. To avoid the particle impoverishment problem of the conventional terrain based navigation, a terrain roughness measure is employed to modify the weight update and resampling steps of the standard particle filter framework. We conducted field experiments in an inland water environment using the designed surface vehicle, and validated enhanced tracking performance of the proposed methods by comparing the methods with conventional approaches.