Mobility modeling for UAV swarm search based on pheromone traces

Abstract

Recently, rapid development of Unmanned Aerial Vehicle (UAV) in a variety of industrial and research field has brought tremendous needs for using a fleet of UAVs to overcome the shortcomings from a use of single UAV. Especially, a fleet of UAVs, so-called UAV swarm, has shown high efficiency in missions that have a huge mission area such as missing person searching, fire forest surveillance and crop field management. However, it needs decentralized and diverse control model to make autonomous UAV swarm accomplish various missions. Because the problem of optimization of searching and surveillance efficiency to control a group of agents usually have high complexity, many researchers tried to adapt nature inspired mobility control model. In particular, the heuristic imitating the way ants use a pheromone as a channel of indirect information exchange has been widely researched. In this thesis, we suggest the improvement algorithm for the pheromone based mobility model by exploiting the neighbor UAVs information. Our algorithm calculate the number of neighbor UAVs in its communication range in real time, and guess what the density near the UAVs is, and then decide whether to go through the previous direction or to change the direction using a pheromone based method. And we also assess our proposed model compared to others such as Random Direction model and basic Pheromone based mobility model on the synthetic mission maps.