Hybrid metaheuristic-driven 3D path planning for UAVs in complex urban environments: a multi-objective fusion framework

Abstract

Unmanned Aerial Vehicles (UAVs) offer transformative potential for urban logistics, but their deployment is challenged by complex 3D path planning in constrained environments. This study introduces GS-PSO, a novel multi-stage hybrid framework designed to address this challenge by sequentially integrating Particle Swarm Optimization (PSO), Sparrow Search Algorithm (SSA), and a Genetic Algorithm (GA). Key innovations include dynamic chaotic inertia weights in PSO for enhanced exploration and specialized GA operators for fine-grained refinement. The efficacy of GS-PSO was validated on the CEC2020 benchmarks, where it found the global optimum in 8/10 functions, showing superior accuracy and stability over standard PSO. In high-fidelity UAV simulations, GS-PSO generated paths up to 22.9% shorter than the AVOA, significantly outperforming competitors. Statistical tests (p < 0.05) confirmed its superior solution quality and robustness. This work provides an effective, reliable solution for complex UAV trajectory planning.