Genetic programming: an introduction: on the automatic evolution of computer programs and its applications
Reflections on the Nature of Multi-Agent Coordination and Its Implications for an Agent Architecture
Autonomous Agents and Multi-Agent Systems
Introduction to Evolutionary Computing
Introduction to Evolutionary Computing
Cooperative Multi-Agent Learning: The State of the Art
Autonomous Agents and Multi-Agent Systems
Hierarchical multi-agent reinforcement learning
Autonomous Agents and Multi-Agent Systems
On a Paradox of Traffic Planning
Transportation Science
Evolving teamwork and coordination with genetic programming
GECCO '96 Proceedings of the 1st annual conference on Genetic and evolutionary computation
Hi-index | 0.00 |
In the wild groups of hyenas are often observed to cooperate in driving lions away from a kill in order to claim it for themselves. Because lions are much larger and can easily injure or kill a single hyena, this is a difficult, high-risk, high-reward behavior requiring complex cooperation by the hyenas and depends on their ability to use communication to coordinate their attack. In this paper we attempt to evolve hyena behaviors that successfully drive-off simulated lions. We are particularly interested in how the type of communication influences the evolution of successful strategies. Several forms of communication are tested including two inspired by hyena behavior. The first is a generalized vocalization or "call" which can be either local or global and used by any hyena. The second mimics the recognition of a special hyena, which we refer to as the "flag-bearer". Our results show that the presence of a flag-bearer leads to the evolution of significantly more effective coordination than either purely local or name based communication. These results suggest that there may be a "sweet spot" between too little information, which makes coordination difficult, and too much information which makes both evolutionary learning and coordination difficult.