High-precision Underwater 3D Mapping Using Imaging Sonar for Navigation of Autonomous Underwater Vehicle

This study proposes a method for generating a high-precision three-dimensional (3D) map using two-dimensional (2D) sonar images from an imaging sonar installed on an autonomous underwater vehicle (AUV). The 2D sonar image sequence was analyzed pairwise to estimate the amount of displacement and used to create a 2D mosaic sonar image. The mosaic sonar map contains intensity information in a wide area and precise shape information but has no height information. To overcome this limitation, we can generate a 3D point cloud from 2D sonar image sequences. This method takes advantage of the mobility of the AUV to reconstruct the height information and partially solves the ambiguity issues in the imaging sonar's elevation angle. The height map generated from the 3D point cloud contains height information of a wide area, but the shape information is inaccurate. By fusing two maps to complement each other's imperfections, we can generate a precise 3D sonar map. This map enables the AUV to estimate the pose and recognize the surrounding environment. We verified the proposed method by conducting experiments in an indoor water tank. After placing various objects on the floor, the AUV with the imaging sonar scanned the floor and objects to generate a 3D sonar map. We analyzed the estimated AUV trajectory and the accuracy of the 3D sonar map.

Automated summary

This study presents a method for generating a high-precision 3D map using sonar images from an autonomous underwater vehicle. By analyzing a sequence of 2D sonar images, a 3D point cloud can be generated to overcome the lack of height information.