Optimal strategy for seabed 3D mapping of AUV based on imaging sonar

An imaging sonar loses information of elevation angle while a mapping process. To overcome this limitation, the motion of the autonomous underwater vehicle (AUV) can be used to obtain 3D information using the imaging sonar. In this paper, we propose a two-stage mapping strategy for accurately generating underwater 3D maps based on an imaging sonar. It consists of searching and scanning stage. In the scanning stage, multi-directional scanning is performed on an object. To process 3D point cloud data obtained by multi-directional scanning, we propose a polygonal approximation method. This method reduces the uncertainty of 3D point cloud data by extracting intersection area of multiple data groups. To verify the feasibility of proposed strategies, we conducted indoor tank experiments using a hovering-type AUV ‘Cyclops’ and acoustic lens-based multibeam sonar (ALMS) ‘DIDSON’.1