Sensor Fusion-based 3D Reconstruction by Two Sonar Devices for Seabed Mapping

Abstract

We propose a sequential method to extract height information for 3D mapping by using sensor fusion of two sonar devices. The 3D reconstruction using sonar image from forward scanning sonar (FSS) is an ill-posed problem, thus we need additional constraints to extract 3D information from the sonar images. Our approach is to utilize complementary information from an additional sonar which is profiling sonar (PS) and geometric constraints between the installations of two sonars. Two sonars are fixed in the specialized installation, thus data from one sonar device can be converted into that of another one by using geometric constraints. The fusion process proceeds in three stages. The first stage is to carve out no object area using the occupancy grid mapping scheme. The second stage is to predict the elevation angle of the FSS by using the PS measurement and geometric constraints. The third stage is particle filtering using the generated particles by computing the weight and to collect the data of FSS. To verify the proposed method, we applied the proposed method in simulation and experiment in engineering basin with AUV.