An Energy-Aware Real-Time Search Approach for Cooperative Patrolling Missions with Multi-UAVs

This paper proposes a novel energy-aware approach to solve the patrolling problem using multi-UAVs (Unmanned Aerial Vehicles), called NC-Drone, which extends the Node Counting method aiming to reduce the number of turns in order to save energy. We present two versions of NC-Drone: a Centralized version and a Decentralized one. The Centralized version collects/spreads the information, reading/writing pheromones in the scenario and it is compared to the real-time search methods. Our results point out that the Centralized NC-Drone overcomes all real-time search methods in most of the performance metrics, presenting better results in Quadratic Mean of the Intervals and drastically reducing the Number of 90° Turning Maneuvers, and thus impacting positively on energy consumption. The Decentralized NC-Drone version uses individual pheromone matrices to guide the vehicles, replacing the centralized scheme of pheromones disposed in the scenario. When a vehicle is close to another, they send its matrices to each other, simulating the use of a short-range communication technology. Additional experiments compare variations of the Decentralized NC-Drone to analyze the impact of the adopted information merge scheme on coverage. The Decentralized version emerges as a promising solution to support real-world applications using low-cost aerial vehicles.