Some Interactions of Information and Control in Integrated Automation Systems

The ready availability of inexpensive and standard communications hardware in the late 1980s, coupled with sufficient standardization of intermediate-layer protocols to make reliable machine-to-machine communication through any combination of standard devices a reality, has laid the groundwork for integrated automation on a grand scale. The contribution of the Manufacturing Automation Protocols (MAP) effort [MAP87], thus far, has been to speed delivery of this powerful communications capability to the industrial community. We must realize, however, that the delivery of the capability has now outstripped our ability to use it. On the factory floor, we have no devices which are prepared to control automation on a grand scale, and even worse, we have few controllers which are capable of contributing their local automation to any kind of integrated whole. The stumbling block is the lack of common languages or even a common experience base in which to communicate about automation tasks. We are now in the midst of a flurry of intellectual activity, on both sides of the Atlantic, developing languages, architectures, control techniques, data systems, communications systems and ultimately controllers to meet this challenge. In such a burst of activity, it is to be expected that experts in individual areas will hasten to solve the problems with which they are most conversant, and that certain interdisciplinary problems will fail to get timely attention. It is the purpose of this paper to draw attention to several such issues now, in the hope that they may get due consideration as the process advances. The issues discussed here are: separation of data flow from control flow,the impact of perfect communication on real-time control,consistency of data with physical reality,the significance of standard data models,the identification of shared and private data.