Feature-based UKF-SLAM Using Imaging Sonar in Underwater Structured Environment

Abstract

This paper presents a simultaneous localization and mapping(SLAM) algorithm towards underwater structured environment using Mechanical Scanning Imaging Sonar(MSIS). An adaptive Hough transform integrating with the method of Random Sampling Consensus(RANSAC) is used to extract the line feature form sonar scanning data and build the geometric feature map in this paper. The UKF-SLAM algorithm estimates the state of underwater vehicle’s pose by fusion of multi-sensor data and the extracted line feature. To validate the algorithm, a simulation on MATLAB using Spanish abandoned marina dataset is tested, which shows this algorithm can suppress the divergence effectively and locate the vehicle accurately.