Drone Surveillance System Considering Dynamic POIs

This study proposes a new drone surveillance system by considering the relative importance of spatial area under the surveillance. Traditionally, the drone surveillance systems employed a simple routing rule often realized as zig-zag movements without alternation of altitude. Recognizing that the altitude of drones can be manipulated to take more detailed pictures of the POIs (Points of Interest), we represent a new mathematical formulation and an efficient heuristic algorithm that considers not only the locations but the altitude of drones. We also incorporate a machine learning algorithm to identify POIs inside the given area, which enables us to implement a fully automated drone surveillance system. The proposed system is superior to the previous ones because the consideration of the relative importance of POIs makes the drone cover the non-POI areas much more efficiently by using higher altitude and taking fewer photos. We implement a proof-of-the-concept of the proposed algorithm, and report results obtained from real-world test cases.

Journal of Korean Institute of Industrial Engineers, 45(5), 416-429, 2019