Decentralized Hybrid Control Strategies for Autonomous Multi-Agent Swarms

"The effective coordination of large groups, or "swarms" of autonomous vehiclesworking collaboratively demands the development of control architectures thatemerge collective intelligence among groups of individuals. Nature, throughevolution and natural selection, has optimized this behavior. Insect societies,in particular, demonstrate an organized "swarm intelligence" beyond the capacityof any individual within their troupe to understand. Although possession of similar capabilities is vital to synergize the performance of multi-agent teams for militarymissions, swarm behaviors cannot be predicted, only observed; resources are needed toevaluate interactions between the entities found in such force structures.In past work, our research group has generated swarm intelligence algorithms mirroringthe capacity of societal insects to emerge collective intelligence. Furthermore, theyhave been successfully interfaced to fabricate a flexible, software system, and a globallyoptimal multi-agent task assignment algorithm. Orbital Research proposes extending thiswork to develop sets of control algorithms that may be configured to direct any swarm ofautonomous agents. Phase I work will: 1) create a multi-agent simulation environment,2) develop dynamic (recursive) control structures for on-line reconfiguration, 3) developcontrol structures to enable optimal prioritization for agents within swarms, 4) simulatethe developed strategies for a candidat