Path planning for unmanned surface vehicles in complex maritime environments via an adaptive improved GA–PSO hybrid algorithm

Unmanned surface vehicle (USV) path planning in coastal and maritime environments remains challenging because of irregular coastlines, densely distributed obstacles, kinematic restrictions, and uncertain environmental threats. This study develops an adaptive improved GA-PSO hybrid algorithm to address these difficulties. The algorithm adjusts inertia weight, crossover probability, and mutation rate adaptively during the search process, while elitism retention and a lightweight local refinement procedure maintain an effective balance between global exploration and local exploitation. Feasibility is preserved through a feasibility handling mechanism that enforces kinematic limits, shoreline and obstacle clearance, and terminal constraints within a consistent WGS-84 to ENU geodetic framework. Path quality is assessed using a multi-criteria objective incorporating length, smoothness, collision penalties, and a continuous-time Markov chain–based threat-exposure metric that characterizes dynamic environmental risks. Experiments using both synthetic scenarios and real nautical-chart data show that the proposed method yields shorter, smoother, and safer routes with consistently lower aggregated costs than GA, PSO, and standard GAPSO, demonstrating enhanced convergence robustness and strong practical applicability for USV navigation in complex coastal waters.