摘要：AbstractAutonomous underwater vehicles (AUVs) are an outstanding tool for underwater exploration. Due to the characteristics of the marine environment that limit the sensors capability, autonomous navigation is still a challenge. This paper proposes a loop closure detector addressed to the Simultaneous Localization and Mapping (SLAM) problem using acoustic images of forward-looking sonars. It is presented an acoustic image descriptor that uses a graph to represent the topologic relationship between the observed objects, a graph comparison approach to determine the similarity between acoustic images and also a segmentation method that uses an adaptative threshold strategy based on the acoustic beams analysis. The method is evaluated using a real dataset of a marina.
关键词：KeywordsMarine system navigationguidancecontrolAutonomous underwater vehiclesMarine system identificationmodellingLocalizationRobot NavigationSLAM
