Implementation of Point Cloud Algorithm on Three Different Installations of Mechanically Scanning Imaging Sonar for Seabed Mapping

摘要

Underwater is very harsh environment in robotics. A reliable localization is one of the fundamental problems for autonomous underwater vehicles (AUVs) because GPS does not work and cameras are limited in use due to the absorption of light. Therefore, sonar is a feasible sensor for underwater localization, which is robust to water turbidity and low illuminance. In this paper, we implemented a point cloud algorithm for mechanically scanning imaging sonar (MSIS) on an AUV for 3D seabed mapping. According to the MSIS installation method, the point cloud algorithm was modified and the results was compared. we considered 3 different installations: 1) tilting forwardly, 2) laying down sideway, and 3) panning in the laid MSIS down. Among 3 installations, case 3 resulted in the adequate. This results will be utilized to investigate a feasible method for the fusion of the other sonar which has complementary characteristics.