P. Stone and M. Veloso
Computer Science Department
Carnegie Mellon University
Pittsburgh, PA 15213
{pstone,veloso}@cs.cmu.edu
Submitted to IEEE Transactions on Knowledge and Data Engineering (TKDE). June 1996.
Updated version in both html and postscript at http://www.cs.cmu.edu/~pstone/pstone-papers.html
Citable as Carnegie Mellon University CS technical report
number CMU-CS-97-193. December, 1997.
Distributed Artificial Intelligence (DAI) has existed as a subfield of
AI for less than two decades. DAI is concerned with systems that
consist of multiple independent entities that interact in a domain.
Traditionally, DAI has focussed on the information management aspects
of these systems. But in the past few years, a subfield of DAI
focussing on behavior management, as opposed to information
management, has emerged. This young subfield is called Multiagent
Systems (MAS). This survey of MAS is intended to serve as an
introduction to the field and as an organizational framework. It
contains guidelines for when and how MAS should be used to build
complex systems. A series of increasingly complex general multiagent
scenarios are presented. For each scenario, the issues that arise are
described along with a sampling of the techniques that exist to deal
with them. The presented techniques are not exhaustive, but they
highlight how multiagent systems can be and have been used to build
complex systems. When options exist, the techniques presented are
biased towards Machine Learning approaches. Additional opportunities
for applying Machine Learning to MAS are highlighted and robotic
soccer is presented as an appropriate testbed for MAS.
Index terms: Multiagent Systems, survey, Machine
Learning, robotic soccer, intelligent agents, pursuit domain,
homogeneous agents, heterogeneous agents, communicating agents