Decentralized Multi-Drone Coordination for Wildlife Video Acquisition

One of the best sources of information for biologists and ethologists to study wildlife behavior is video footage; in particular, aerial video footage provides a unique perspective on the behavior of animals in their natural habitat. Numerous wildlife behavioral studies have demonstrated the effectiveness of UAVs for collecting video data of group-living animals. However, in contrast with well-established techniques for static video acquisition, the deployment of UAVs for wildlife video acquisition requires human operators to manually control and coordinate the drones while minimizing disturbance to animals. To scale UAVs missions to obtain sufficient spatiotemporal resolution, reliance on manual operations is impractical. In this paper, we present a decentralized multi-drone coordination system for wildlife video acquisition using UAVs that leverages a novel k-coverage algorithm specifically designed to cover herds. In particular, it is based on a hierarchical clustering approach to find the herds, centroids, then it coordinates multiple drones in a decentralized fashion to cover them from multiple points of view. We introduce a set of metrics to evaluate the effectiveness of the proposed approach via simulation, finding that the proposed approach improves noticeably over the present state of the art.