The Impact of Heterogeneity on Cooperating Agents

The majority of previous work in Distributed AI (DAI) has concentrated on cooperation between similar agents. Such agents typically have homogeneous structures (e.g. BEINGS [Lenat75] or DVMT [Lesser&80]), are capable of carrying out identical tasks (eg Air Traffic Control [Cammarata&83]) or are assumed to have identical domains of discourse. However such cosy assumptions and the way they simplify the problems of coordination and cooperation, are often not applicable in real-world situations real problems involve heterogeneity! Within the ARCHON1 project we are developing a cooperation environment for industrial systems. There are two main characteristics of this domain which impact upon the design of our system, both of which need to be tackled if DAI is ever to leave the laboratory phase and progress into the “real” world: