LOTTO: group formation by overhearing in large teams

  • Authors:
  • Francois Legras;Catherine Tessier

  • Affiliations:
  • Onera DCSD / SupAéro, Toulouse, France;Onera DCSD, Toulouse, France

  • Venue:
  • AAMAS '03 Proceedings of the second international joint conference on Autonomous agents and multiagent systems
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

We present in this paper an extension to our overhearing-based group formation framework \otto (Organizing Teams Through Overhearing). This framework allows a team to dynamically re-organize itself in accordance to the evolution of the communication possibilities between agents. In \otto, each agent overhears some of the messages exchanged within the team and uses them to incrementally update a map of the organization of the team. One of the key points of \otto is that only few resources are used - a small data structure is used to track each team member and there is no need to store the overheard messages. With \lotto (\otto for Large numbers of agents) we address the problem of large teams in which using \otto "as is" would be costly in terms of memory and therefore contrary to \otto's philosophy of low resource usage. Therefore, we have implemented a strategy that allows an agent with a limited memory to track only a part of the team. We have run a series of experiments in order to evaluate the impact of this limitation and present some results.