Research On Ant Colony Algorithm For Self-Organization Of UAVs' Mission In Unknown Environment

Unmanned aerial vehicles(UAVs) attract more and more countries because of their predominance in modern wars. With the increasing complication of the battlefield, the increasing diversity of the military mission and the increasing possibility of multiple UAVs performing a certain military mission cooperatively, it is not suitable for the coming war that UAV is controlled by mission control state(MCS) completely. For this reason, it is necessary to investigate methods of making UAVs campaign autonomously based on existing vehicles.This dissertation does researches on how to improve the capability of campaigning autonomously of multiple cooperative UAVs in future unknown battlefield effectively with self-organized algorithm by imitating the behavior of ants faring. The main work and creative contribution of this thesis are as follow:(1) The self-organization characteristics of basic ant colony algorithm(ACA) is discussed. According to the definition of self-organization and the mechanism of positive/negative feedback, the self-organized characteristics of ACA, such as the stigmercy of swarms and interaction among units etc. And the behavior of ants mustered by the food found is revealed from the view of problem solving.(2) A distributed vector structure of pheromone is proposed. According to the distributed control infrastructure of UAVs, a distributed structure of pheromone is designed. As a result, it will improve the robustness of the self-organized algorithm designed in this thesis, and reduce the dependence on the condition of communication. In order to overcome the problem that traditional scalar pheromone only reflect simplex information of problem solving, a vector pheromone is designed. It will be consistent with the characteristics of multi-kind UAVs performing various mission. Simulation results demonstrate that the algorithm holds robustness to the condition of communication and the ants carry on various action reasonably under the distributed vector structure of pheromone.(3) A new mechanism of updating pheromone based on reactive action is presented. Based on ACA, a new mechanism of updating pheromone, which simulates the simple reactive behavior of ants, is designed according to different actions that UAVs perform. And a method of limiting the extent of pheromone changing dynamically is introduced into the mechanism, so that the UAVs will take sustaining action over targets. Simulation results demonstrate that UAVs will head to maximize the goal of military mission, and perform sustaining action over targets under the mechanism.(4) Method of making UAVs be cooperative based on pheromone views is researched. Based on the mechanism that the behavior of ants is inducted by pheromone, a policy of exchanging information among UAVs is proposed to direct pheromone updating. It will reduce the traffic and the frequency of communication, at the same time it will keep the pheromone view reflect the situation of battlefield effectively. A mechanism of state estimation, which includes a forward-task-compensation policy and a backward-state-estimation policy, is designed to reduce the impact of communication on the consistency of pheromone views in different ants. With the mechanism, the cooperation among UAVs will be enhanced, and the campaign efficiency of UAVs will be improved. Simulation results show that the impact of communication on cooperation among UAVs is reduced, and the efficiency of the cooperation is enhanced.