Robust AUV Localization Based on Switchable Constraints

In mission of marine exploration, AUV is often used to perform seabed exploration and map construction on the sea floor. As there are many similar sea-floor images, AUV often suffers false positive loop closures in similar underwater environments during Simultaneous Localization and Mapping (SLAM). In this paper, a robust optimization method is proposed to eliminate loop closures during AUV localization and mapping, based on switchable constraints. The proposed method is verified both qualitatively and quantitatively in simulation and underwater experiments.