Judging the value of additional information on the performance of intelligent agents in manufacturing control

In recent years we have seen the exploration of the use of multiagent systems to provide control of manufacturing systems. This paper provides a brief review of agent-based systems applied to manufacturing control discussing the principle factors that contribute to defining the uniqueness of their design. The question arises in the de development of such systems as to how much information about present processing and predicted future processing should be used in the negotiation and generation of plans for use in routing parts through the system. This paper explores this issue by first discussing the details of the negotiation mechanism employed. Then the results of tests are presented that compare the performance of the manufacturing system when the controlling agents are provided with the different forms of information.