Scalable real-time multi-UAV coverage path planning for large geographical areas

Multi-Uav systems enable rapid wide-area coverage and monitoring, but mission planning for large fleets over very large areas is complex and slow. In this paper, we present a scalable real-time strategy for large geographical areas that couples centralised partitioning with decentralised planning. The method integrates k-means clustering, Voronoi tessellation, and lawnmower patterns. We demonstrate it on diverse real-world areas, including the largest national parks in the UK and the U.S., Quebec, and Antarctica; and analyse its variability to highlight the algorithm’s flexibility and robustness. Finally, we show the system’s effectiveness and scalability by planning a near-complete coverage of the state of California in 10 h with 40 UAVs cruising at 32 m/s (115 km/h).