AUTONOMOUS NAVIGATION USING SLAM AND A MARITIME RADAR SYSTEM

Abstract

This paper presents the design and evaluation of an Unmanned Surface Vessel (USV) capable of autonomous navigation and remote coastline mapping. The system implements a radar-based solution with a Simultaneous Localization and Mapping (SLAM) algorithm to construct large- scale ground-truth maps for both localization and path planning. Radar is leveraged over LiDAR for long-range scanning due to its superior range and higher scan accuracy. The SLAM algorithm works by extracting and matching features from multiple radar scans, which allows it to stitch new features onto a large-scale generated map. The generated map is used in tandem with the on-board LiDAR unit, with the map being used to plot out proposed courses for the USV, while the LiDAR employs basic obstacle detection and avoidance techniques while pathing to the desired location.