Decentralized Path Planning Approach for Crowd Surveillance using Drones

The problem of periodical surveillance of multiple moving targets using multiple drones in a decentralized manner is addressed in this paper. The strategy proposes a traverse order generation scheme by which the sequence in which targets are tracked can be determined. To cover all the targets by visiting them sequentially, the predicted targets' future position is used in path planning. The proposed algorithm takes care of drones' assignment to targets and eventual hand-off to neighboring drones as targets move from one region to another. The proposed strategy is used for a crowd monitoring system, and it attempts to detect the anomalous activity of a crowd via target trajectories. Complete surveillance architecture is validated using simulation for tracking of multiple walking agents using multi drones.