LiDAR-Based Autonomous Docking System for Uncooperative Vessels in GNSS-Denied Environments Using a Size-Adaptive Dubins Path

Docking an uncrewed surface vehicle (USV) to a target is a complex decision-making task that requires cooperative autonomous perception and navigation, becoming significantly more challenging in real marine environments where standard localization methods, such as the global positioning system, are unavailable or disrupted. In this article, we design a light detection and ranging (LiDAR)-based perception-planning autonomous system for a USV to dock to an uncooperative vessel without using Global Navigation Satellite System signal, where the uncooperative vessel is defined to be anchored yet subject to swing or drift due to waves or currents, with no communication or pose sharing except providing two proper sides for docking. To achieve successful docking, we first employ a clustering algorithm and L-shape fitting for onboard LiDAR-based object detection to locate the target and estimate its heading. Next, we develop a perception-integrated Dubins path planning method to generate a smooth trajectory around the target, enabling precise heading estimation by the LiDARs for final docking. Finally, we incorporate a station-keeping mechanism to ensure stable attachment to the target, forming a comprehensive solution. Real-world experiments under sea state three conditions confirm the robustness of this approach.