Experiments on localization of an AUV using graph-based SLAM

This paper presents experiments on vision-based localization of an autonomous underwater vehicle (AUV) using graph-based simultaneous localization and mapping (SLAM). Relative range and bearing values of each landmark are obtained from image processing results. And a graph structure is built using the landmark detection results and dead-reckoning data of the AUV. The structured graph is optimized by a graph-based SLAM algorithm. Finally, the performance of the graph-based SLAM is compared to an EKF-based SLAM result.