Greedy dynamic reward algorithm-based coverage path planning for unmanned sailboats in non-stationary wind environments

Abstract

Unmanned sailboats, powered by wind, are well-suited for long-term island patrols but face challenges in maneuverability due to wind constraints, limiting route flexibility. This study addresses the issue of coverage path planning by proposing a Greedy Dynamic Reward Algorithm that incorporates wind conditions and sailboat kinematic constraints. A sailboat movement model is developed using a velocity prediction program. With dynamic rewards assigned to map grids based on wind direction, trajectory, and obstacles, the sailboat selects the grid with the highest reward among the available options for its next move. In cases of deadlock, a Breadth-First Search algorithm identifies the nearest uncovered node, and an improved artificial potential field method is employed to navigate to that node. Key strategies, including occupying the relatively most upwind position and disabling movement directions based on the sailed path, effectively minimize overlap and enhance the trackability of the path. Simulation results demonstrate the algorithm's robustness and adaptability in both single-island and archipelago environments, even under changing wind conditions. Furthermore, the study discusses potential directions for algorithm optimization.