Distributed Control of a Swarm of Autonomous Unmanned Aerial Vehicles

摘要

With the increasing use of Unmanned Aerial Vehicles (UAV)s military operations, there is a growing need to develop new methods of control and navigation for these vehicles. This investigation proposes the use of an adaptive swarming algorithm that utilizes local state information to influence the overall behavior of each individual agent in the swarm based upon the agent's current position in the battlespace. In order to investigate the ability of this algorithm to control UAVs in a cooperative manner, a swarm architecture is developed that allows for on-line modification of basic rules. Adaptation is achieved by using a set of behavior coefficients that define the weight at which each of four basic rules is asserted in an individual based upon local state information. An Evolutionary Strategy (ES) is employed to create initial metrics of behavior coefficients. Using this technique, three distinct emergent swarm behaviors are evolved, and each behavior is investigated in terms of the ability of the adaptive swarming algorithm to achieve the desired emergent behavior by modifying the simple rules of each agent. Finally. each of the three behaviors is analyzed visually using a graphical representation of the simulation, and numerically, using a set of metrics developed for this investigation.