Strategies for Patrolling Missions with Multiple UAVs

This paper proposes a set of strategies for the patrolling problem using multiple UAVs and as a result, improving our original NC-Drone algorithm. We present four strategies: Watershed Strategy, Time-based Strategies, Evaporation Strategy, and Communication-Frequency Strategy. The novel strategies consider important aspects of the patrolling movement, such as time, uncertainty, and communication. Results point out that these strategies improve the centralized version of the NC-Drone considering the uniform distribution of visits and drastically reduce in 76% the standard deviation, making the algorithm more stable. Based on the results, we found that there is a trade-off between the evaluated metrics, making it necessary to perform a large number of turns to obtain a more spatially distributed patrolling. We also present a series of strategy combinations, achieving slight improvements as more combinations are adopted. The resulting algorithm from the combination of all strategies reduces the communication frequency in 50 times and outperforms the original version of the NC-Drone in 4.5%.