Abstract: Persistent surveillance is a typical application of multi-swarm aerial vehicle systems (UAVs). And dynamic deployment for multi-swarm UAVs in persistent surveillance has been proved to be a complex problem, especially when the self-adjustment is required to adapt the time-sensitive environment. This paper proposes a multi-swarm hierarchical control scheme and key algorithms. We design the digital turf potential field model to approximate the evolving and interactive information of time-sensitive target features and surveillance effects. Moreover, using the digital turf potential function of each grid as the data point weight, we design a grid-based weighted data-clustering algorithm for the dynamic assignment of UAV swarms, which can adaptively adjust the number of UAVs in each swarm and its sub-region. Finally, we evaluate the proposed architecture by means of case studies and find that our method can promote surveillance efficiency and workload balance of multiple UAV swarms.

Key words: multi-UAV swarm, persistent surveillance, dynamic deployment, digital turf model, data clustering