Graph-Based Bathymetry Matching for Robust Localization

Beomsoo Kim, Sanghoon Lee*

Terrain-based localization using bathymetric maps has emerged as a promising solution in Global Positioning System (GPS) denied environments for autonomous vehicle. However, conventional methods that directly compare bathymetric data often suffer from ambiguity caused by similar terrain patterns so called local minima, vulnerability to sensor noise, and high computational complexity. This paper presents a novel graph-based matching approach that transforms bathymetric data into a structural graph representation for robust localization. Key terrain features are extracted as nodes, and their geometric relationships are expressed as edges, enabling reliable matching even under partial observations and noisy conditions. Experimental results demonstrate that the proposed method successfully estimates location in scenarios where Iterative Closet Point (ICP) matching methods fail, while achieving higher overall matching accuracy and computational efficiency. This study provides an effective and robust bathymetry- based localization solution for environments where GPS signals are unreliable or unavailable.

Keywords: underwater terrain matching, registration, graph-matching