Flow-aware hierarchical planning and execution for USV inspection in offshore wind farms

Offshore wind farm inspection by unmanned surface vehicles (USVs) requires coordinated target visitation, service-constrained turbine entry, current-aware motion planning, and reliable mission execution. This paper presents a flow-aware hierarchical framework in which candidate entry poses connect task sequencing with inspection service requirements. Current-informed transition costs jointly determine turbine order and entry selection, while guided Hybrid A* search, online entry adaptation, and strict-first recovery support local execution. The framework is evaluated in reconstructed Beatrice and synthetic complex-flow scenarios. To connect the representative vessel dynamics with measured behaviour, the adopted three-degree-of-freedom (3-DOF) model is assessed against published WAM-V USV14 field measurements. The straight-acceleration comparison gives a surge-speed root-mean-square error (RMSE) of 0.131 m/s, while the simulated maximum absolute yaw rate is 10.13°/s compared with 13.07°/s in the digitized differential-thrust zig-zag measurement. Using the same model without field-data-based retuning, the frozen 46.80-km Beatrice trajectory achieves a cross-track RMSE of 4.90 m, completes all eight dwell poses and six service geometries, and maintains a 12.17 m dynamic safety margin. Current-informed sequencing reduces mission time, the planning-level propulsion-energy indicator, and the total objective by 5.3%, 8.7%, and 5.3%, respectively.