HMA-RRT*: A hybrid multi-strategy adaptive RRT* algorithm for USV path planning in complex maritime environments

Abstract To address the global path planning challenge for Unmanned Surface Vehicles in complex maritime environments characterized by dense islands and narrow waterways, this paper proposes a Hybrid Multi-Strategy Adaptive RRT* algorithm. The method combines a dynamic region-based sampling strategy with an improved artificial potential field based dynamic extension strategy, which introduces random-node attraction, dynamic repulsion adjustments, and additional repulsive forces. Additionally, a hierarchical side-retreat escape mechanism is applied to enhance obstacle avoidance and search efficiency in complex environments. The algorithm also incorporates heading-angle constraints and adaptive step-size adjustment to ensure the path complies with USV kinematic properties. Furthermore, an improved NSGA-II algorithm is proposed to perform multi-objective optimization of path length, smoothness, and safety, and B-spline interpolation is used to generate continuous and executable paths. Simulation results show that, compared with the standard RRT* algorithm, the proposed HMA-RRT* algorithm achieves average reductions of 7.85% in path length, 66.96% in node count, 48.73% in computation time, and 25.7% in mean turning angle across four representative complex maritime environments. These improvements significantly enhance search efficiency, path smoothness, and planning feasibility, thereby providing a reliable and efficient path-planning solution for autonomous USV navigation in complex maritime conditions. Graphical abstract