GNSS-Independent Maritime Navigation Using Monocular Camera Images with Digital Elevation Map

Abstract

The threat of GNSS service disruption implies the need for GNSS-independent navigation solutions. We introduce a framework that estimates the position of a maritime vessel in a littoral zone through monocular camera images and existing knowledge of the vessel’s environment. A set of algorithms extract and match image features from a camera aboard the vessel with synthetic images rendered from a digital elevation model (DEM). The image features, together with the knowledge of the vessel’s intended position, are used to infer a 3-dimensional feature inside the DEM. A motion-only bundle adjustment optimization problem is then posed, seeking to estimate the actual position of the vessel by minimizing the reprojection error of the DEM feature. From 57 independent estimation problems with a variance of 70m, the average error results in approximately 25m, demonstrating the technique as a potential candidate for GNSS-independent position estimation.