Using emergence in participatory simulations to design multi-agent systems

  • Authors:
  • Paul Guyot;Alexis Drogoul;Christian Lemaître

  • Affiliations:
  • Université Pierre et, Paris;IRD Bondy, Bondy Cedex;Mexico

  • Venue:
  • Proceedings of the fourth international joint conference on Autonomous agents and multiagent systems
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

In this paper, we present, through simulations of the coffee market of the state of Veracruz, how emergence of specialized roles in participatory simulations could be used to design and improve multi-agent systems.The design process of the participatory simulation followed the regular design process of multi-agent simulations: starting from a domain model of coalitions among coffee producers, we built a participatory simulation where humans take the control of agents. Special care was brought to make the simulations playable by humans and to favor the apparition of coalitions. Being controlled by humans, the agents of the initial model became completely pro-active and were able to exchange coffee and money and to form coalitions to fulfill the buyer's offers. Coalitions appeared as expected.Besides, during these experiments, we observed the emergence of specialized roles which were not included in the initial model. We implemented a regular multi-agent system based on the initial model to test the distributed system solving improvements brought by the roles that emerged.