Photo Mosaicing of Tagiri Shallow Vent Area by the AUV "Tri-Dog 1" using a SLAM based Navigation Scheme

Abstract

This paper proposes a navigation scheme of an Autonomous Underwater Vehicle (AUV) for photo mosaicing of a shallow vent area with bubble plumes. While bubble plumes disturb acoustic positioning systems, they can be detected by sonars. So the method takes advantage of the plumes as landmarks using a profiling sonar. By adopting the concept of Simultaneous Localization and Mapping (SLAM), the method achieves drift-free, accurate and independent positioning without using conventional acoustic transponders. The high positioning accuracy enables complete data acquisition, as well as position based mosaicing without relying on pictorial correlations of the photos. Some artificial acoustic reflectors are also deployed to enhance positioning performance. The scheme was implemented to the testbed AUV "Tri-Dog 1" and the performance was verified through sea experiments at Tagiri vent area, Kagoshima Bay in Japan. The AUV succeeded in fully autonomous observation at the challenging environment, to build photo mosaics of more than 300 m of the floor including a tube-worm colony