3D Reconstruction Using Two Sonar Devices in a Monte-Carlo Approach for AUV Application

Abstract

We propose a sequential method to extract 3-dimensional information for mapping by using sensor fusion of two sonar devices. The 3D reconstruction using sonar images from forward-looking multibeam sonar (FLMS) is an ill-posed problem due to the loss of elevation angle; thus we used an additional sonar and its complementary information. Our approach is to utilize geometric constraints between the installation of two sonar devices and complementary characteristics such as a different angle of the acoustic beams and data acquisition method. The fusion process proceeds in three stages. The first stage is to build a likelihood map using geometric constraints of the installation between two sonar devices. The second stage is to calculate the feasible elevation angle of the FLMS. The elevation angle of FLMS is limited by its vertical beam spreading angle, which can be represented by particles. The third stage is Monte-Carlo experiment using the generated particles by computing the weight, and correct the data of FLMS. To verify the proposed method, we applied the proposed method in simulations and experiments in engineering basin with an AUV.