Coverage Path Planning for Autonomous Sailboats in Inhomogeneous and Time-Varying Oceans: A Spatiotemporal Optimization Approach

Autonomous sailboats offer favourable endurance for long-duration ocean observation. Their sailing performance is, however, strongly anisotropic with respect to the wind, while wind and current fields are spatially inhomogeneous and temporally varying, so that certain regions may be accessible only from limited directions during particular intervals. Existing coverage path planning (CPP) methods typically rely on static regional decomposition and boustrophedon sweeping, and are therefore ill-suited to performance anisotropy and partial directional inaccessibility; prior point-to-point sailboat planners likewise do not directly address the coverage problem.

To this end, the present study proposes a spatiotemporal CPP framework tailored to these conditions. In the spatial domain, topology-based morphological constraints regulate the geometry of the covered region in order to reduce fragmentation and excessive backtracking. In the temporal domain, forecast-aware look-ahead planning exploits meteorological predictions and balances short-term efficiency against long-term feasibility under the assumption that forecast uncertainty accumulates with prediction horizon. The environment is discretised on a grid, a 16-direction action set is adopted, and effective velocity is obtained by combining a velocity prediction program (VPP) with the ambient current. Simulations in stochastic inhomogeneous and time-varying seas, including cases with partial directional inaccessibility, demonstrate clear improvements in coverage efficiency and path feasibility relative to boustrophedon baselines, and provide a foundation for future cooperative multi-sailboat coverage.