Adaptive Control of Distributed Agents Through Pheromone Techniques and Interactive Visualization

摘要

Software agents guided by synthetic pheromones can imitate the stigmergetic dynamics of insects. The resulting software architecture is well suited to problems such as the control of air campaigns and unmanned robotic vehicles. An architecture for the deployment of synthetic pheromones in a military setting is described. The architecture can be instantiated into a single computer, a distributed network of computers, or even across a mass network of field processors implanted in the environment. Software agents representing the military units of interest move through this synthetic environment sensing and depositing synthetic pheromones. These agents are able to locate and select targets dynamically. Using ant colony optimization techniques, the agents build safe paths to the target that adapt in real time to new threats or opportunities. The results produced can be used to help guide manned missions, or to directly control unmanned robotic vehicles. The technology has particular advantages for controlling large swarms of small, unmanned vehicles.