An anytime Visibility–Voronoi graph-search algorithm for generating robust and feasible unmanned surface vehicle paths

• Published in: Autonomous robots Vol. 46; no. 8; pp. 911 - 927
• Main Authors: Schoener, Marco, Coyle, Eric, Thompson, David
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.12.2022; Springer Nature B.V
• Subjects: Algorithms; Artificial Intelligence; Barriers; Computer Imaging; Control; Cost function; Empirical analysis; Engineering; Environment models; Feasibility; Mechatronics; Obstacle avoidance; Optimization; Path planning; Pattern Recognition and Graphics; Robotics; Robotics and Automation; Safety; Search algorithms; Surface vehicles; System effectiveness; Uncertainty; Unmanned vehicles; Vehicles; Visibility; Vision; Voronoi graphs; USV; Path planning; Feasibility; Anytime algorithm; Graph search
• ISSN: 0929-5593; 1573-7527
• DOI: 10.1007/s10514-022-10056-7

While path planning for Unmanned Surface Vehicles (USVs) is in many ways similar to path planning for ground vehicles, the lack of reliable USV models and significant maritime environmental uncertainties requires an increased focus on robustness and safety. This paper presents a novel graph construction method based on Visibility–Voronoi diagrams that allow users to tune path optimality and path safety while considering vehicle dynamics and model uncertainty. The vehicle state is defined as both a 2D location and heading. The method is based on a roadmap generated from a Visibility–Voronoi diagram, and uses motion curves and path smoothing to ensure path feasibility. The roadmap can then be searched using any graph-search algorithm to return optimal paths subject to a cost function. This paper also shows how to generate and search this roadmap in an anytime fashion, which makes the method suitable for local planning where sensors are used to build a map of the environment in real-time. This approach is demonstrated effectively on underactuated systems, with empirical results from USV docking and obstacle field navigation scenarios. These case studies show the path maintains feasibility subject to a simplified vehicle model, and is able to maximize safety when navigating close to obstacles. Simulation results are also used to analyze algorithm complexity, prove suitability for local planning, and demonstrate the benefits of anytime roadmap generation.