An energy-efficient path planning method for USV under ocean currents based on the Biased-SVF-RRT*

Path planning for Unmanned Surface Vehicles (USVs) in large-scale maritime environments is crucial for enhancing navigation efficiency, ensuring operational safety, and reducing energy consumption. To overcome the limitations of the Stream-based Vector-Field RRT* (SVF-RRT*)–namely, slow initial path generation in upstream regions and weak convergence of heuristic sampling-this study proposes a Biased-Sampling-based Stream Vector-Field RRT* (Biased-SVF-RRT*). The proposed algorithm incorporates a goal-directed extension mechanism to accelerate tree expansion under adverse currents, an adaptive cost-weighting strategy that dynamically balances path length and energy consumption according to environmental features, and a sliding-window path detection mechanism that periodically refines heuristic sampling intervals before the initial path is obtained while integrating a pruning process to enhance search efficiency. After the initial path is generated, a soft-constrained sampling mechanism, based on local path-segment evaluation, is employed to maintain a balance between local optimization and global exploration. Moreover, it is smoothed using a cubic quasi-uniform B-spline curve to ensure controllability and trackability. Simulation results indicate that Biased-SVF-RRT* significantly outperforms existing methods in both planning efficiency and path quality under current-disturbed conditions and exhibits strong robustness in obstacle-rich environments. The sensitivity analysis further confirms its superior overall performance under different exploration efficiency factors and step sizes.