Multi-unmanned aerial vehicle (UAV) collaborative trajectory planning in complex three-dimensional (3D) environments faces critical challenges in real-time coordination, including balancing path optimality, collision avoidance, and computational efficiency under dynamic obstacle fields. While existing metaheuristic algorithms exhibit limitations in convergence speed and adaptability to high-density obstacles ( \( > 0.8 threats/km^{3}\) ), this study introduces PHOENIX—a hybrid metaheuristic framework that synergizes the crested porcupine optimizer (CPO) and ant lion optimizer (ALO) through a dual-layer cooperative mechanism. Three key innovations are proposed: (1) A hierarchical flight protection zone model compliant with RTCA DO-263 standards, dynamically integrating CPO’s collision-avoidance strategies with real-time risk prediction; (2) A chaos-enhanced initialization strategy employing interval-constrained logistic mapping, which increases population diversity by 32.7% (quantified via Shannon entropy, \( p<0.05\) ); (3) An adaptive weight adjustment mechanism with nonlinear decay, achieving 23.5% and 18.7% faster convergence than particle swarm optimization (PSO) and grey wolf optimizer (GWO), respectively. Extensive evaluations on LANDSAT 9-derived 3D terrains demonstrate PHOENIX’s superiority: it reduces average path length by 6.2%, decreases threat exposure by 7.3%, and improves trajectory smoothness by 8.9% compared with state-of-the-art benchmarks. AirSim-based simulations with five UAVs validate 100% collision-free success rates in dynamic urban scenarios, outperforming CPO and ALO in robustness ( \( p<0.01\) ). The framework effectively resolves exploration-exploitation dilemmas while maintaining computational efficiency ( \(<50ms \) per planning cycle), offering a scalable solution for mission-critical applications such as disaster response and smart city logistics. By bridging bio-inspired optimization with complex system dynamics, this work advances intelligent autonomous system research.
