Enhancing AUV path planning via state-aware adaptive particle swarm optimization in complex marine environments

Path planning for Autonomous Underwater Vehicles (AUVs) in complex marine environments requires an optimizer that can account for coupled path-quality and safety criteria while adapting its search behavior to individual particles. This paper proposes Adaptive Elite Particle Swarm Optimization (AEPSO) for three-dimensional marine environments that include seabed terrain, obstacles, and ocean currents. Path length, smoothness, ocean-current resistance, and navigation safety are combined in a unified fitness function. Fitness ranking is used to characterize particle states and tune the parameters of individual particles, while distance-adaptive Gaussian mutation is applied to elite particles to balance exploration and convergence stability. Simulation results show that AEPSO achieves lower mean fitness values than GSPSO, APSO, PSO, ACO, and ABC. In the 1000 m × 1000 m × 400 m environment, AEPSO reduces the fitness value by 8.34% relative to A* and by 42.72% relative to RRT, while producing smoother trajectories with fewer inflection points. • High-fidelity 3D marine environment modeling. A realistic 3D seabed, obstacle, and current model avoids over-simplified settings. • Novel adaptive particle swarm optimization (AEPSO). AEPSO tunes inertia and learning factors and mutates elites, raising fitness by 18%. • Multi-objective constrained path planning with engineering feasibility. Length, smoothness, safety, and energy are balanced via B-spline path smoothing.