Dynamic-robustness-oriented hierarchical task planning for UAV swarm based on improved pigeon-inspired optimization

Unmanned aerial vehicle (UAV) swarm task planning in obstacle-dense environments is a large-scale combinatorial optimization problem with strict constraints on energy, feasibility, and adaptability. To address these challenges, a dynamic-robustness-oriented hierarchical task planning framework based on an improved pigeon-inspired optimization (I-PIO) algorithm is proposed. The planning problem is divided into three layers: task allocation, task sequencing, and path planning. A hybrid clustering strategy that combines K-means and hierarchical clustering is used to balance task distribution and lower computational complexity. Within each subregion, an I-PIO algorithm is applied, incorporating fitness-relative adaptive differential updating, probability-driven differential hybrid search, and progress-based elite tangent perturbation to improve global exploration and late-stage convergence. For path planning, a two-layer scheme based on A* search and Bezier smoothing is adopted to ensure feasibility while minimizing energy consumption. A dynamic re-planning mechanism is also introduced to handle UAV failures and task injections. Simulation results show that the proposed hierarchical task planning framework significantly outperforms traditional approaches in efficiency, robustness, and scalability, highlighting its strong potential for UAV swarm mission planning in complex environments.