Simultaneous Localization and Mapping Using Imaging Sonar for an Autonomous Underwater Vehicle

摘要

To implement completely autonomous behaviors for autonomous underwater vehicle (AUV) in unknown subsea environment, an imaging sonar is herin used as main navigation sensor to actively detect features in environment. A series of improved approaches are applied to process sonar images sad extract terrain point-features, then a modified RBPF method is proposed to use in navigation and localization for our underwater vehicle, C-RANGER, where unscented Kaiman filter(UKF) instead of extended Kalman filter is employed to incorporate the current observed feature points as well as historical observations into the proposal distributton. Simulation experiments show that proposed algorithms of feature extraction and UKF-based SLAM are effective. even in large scale environments.