UAV Swarm Mission Planning Development Using Evolutionary Algorithms - Part I SCI-195

Embedding desired behaviors in autonomous vehicles is a difficult problem at best and in general probably impossible to completely resolve in complex dynamic environments. Future technology demands the deployment of small autonomous vehicles or agents with large-scale decentralized swarming capabilities and associated behaviors. Various techniques inspired by biological self-organized systems as found in forging insects and flocking birds, revolve around control approaches that have simple localized rule sets that generate some of the desired emergent behaviors. To computationally develop such a system,an underlying organizational structure or framework is required to control agent rule execution. Thus, autonomous self-organization features are identified and coalesced into an entangled-hierarchical framework. The use of this self-organizing multi-objective evolutionary algorithmic approach dynamically determines the proper weighting and control parameters providing highly dynamic swarming behavior. The system is extensively evaluated with swarms of heterogeneous vehicles in a distributed simulation system with animated graphics. Statistical measurements and observations indicate that bio-inspired techniques integrated with an entangled self-organizing framework can provide desired dynamic swarming behaviors in complex dynamic environments.