A developmental approach to evolving scalable hierarchies for multi-agent swarms
Proceedings of the 12th annual conference companion on Genetic and evolutionary computation
Hi-index | 0.00 |
Hierarchical control structures for multi-agent systems represent a promising middle ground between fully-distributed systems and centralized control. In this paper we present a developmental approach for evolving hierarchical control structures for large (100-800 member), multi-agent swarms. The results show that this approach can successfully generate control hierarchies that improve the performance of fully distributed swarms and that scale well.