Terrain based localization for pinpoint observation of deep seafloors

Real-time accurate localization is essential for in many underwater vehicle applications. In particular, close-range seafloor observation requires precise positioning in relation to bottom structures or to coordinates on a map. This paper proposes a real-time terrain based localization method that realizes robust positioning that does not require any artificial landmarks. The method stochastically determines vehicle´s position relative to the map using the Particle Filter method, based on an inertial navigation system (INS) and a profiling sonar mounted on the vehicle. The reference bathymetric map is constructed a prior, either by the vehicle itself, or by other platforms. The proposed method is implemented on the hovering type AUV TUNA-SAND and sea trials were carried out. We succeeded in the real-time localization and verified the performance of the proposed method.