Multi-Agent Decision Making Architecture and Distributed Control

Abstract

The paper presents principles behind a multi-agent system DISCIM supporting distributed decision making. Using analogical ideas, a distributed architecture for control engineering purposes has been proposed. Each node (agent) in this architecture is expected to consist of two parts: a stand-alone control device and a wrapper. Each part runs on an independent hardware (processor). Standardized communication interfaces/languages are considered. The solution of a reconfigurability problem is described in detail. Brief comparison of the multi-agent systems for decision making and distributed control is presented.