Estimating vehicle position using few range measurements to a known curved surface

Abstract

This paper presents a method by which data from two or more range-finders can be combined to estimate vehicle position when operating in proximity to a surface of known curvature. Estimates are combined in an Extended Kalman Filter which accounts for the expected transformation of sensor noise. Although developed for use with Autonomous Underwater Vehicles, this curvature-based localization technique can be applied to any mobile agent capable of measuring its range to a previously-mapped surface. © 2008 Springer-Verlag Berlin Heidelberg.