How does the subsystem interact with other subsystems on the aircraft, especially with regard to the previous questions?
When would using the subsystem be inappropriate or erroneous (as opposed to ineffectual or irrelevant)?
It is widely suspected that those who construct new systems do not fully understand all the ramifications and implications of what they are designing. Answering these questions will challenge the designers of traditional avionics.
Those who have participated in the design of an intelligent interface have found that the scrutiny given the traditional avionics design can produce a more purposeful product. During design, a number of intelligent interface models are constructed of how the entire system will be used from the pilot's perspective. This model building can yield benefits by improving the design as well as incorporating the intelligence interface functionality. For example, I was once preparing the knowledge base for an information manager that was to select from one of several available displays. It used information requirements that were associated with intention structures, and it picked the best display by matching its information display capabilities to the information requirements that had been accumulated from all active intentions.
While I was debugging the knowledge base, I noticed that some displays were never chosen and that other displays were frequently chosen. Naturally, this was assumed to be a fault of the knowledge base, as it was under development. After close observation of the display selection algorithm, I came to the conclusion that the algorithm and knowledge base were correct. The problem was in the displays themselves. Some displays lacked elements that were always demanded. Other displays seemed to support situations that would never occur. To fix the problem, new display designs were prepared. The point of this example is that evaluation of the information content of displays was made possible only by computing a match of displays to situations. Although it would certainly be possible to prepare a written argument that the displays are well designed, computation was a more compelling proof.
The strength of this approach lies in the executable nature of the knowledge. It is not merely that the knowledge can then be applied via execution to produce simulations of the effects of the subsystems along with the associated knowledge. As such, it represents a powerful system engineering capability that is especially useful to those who are responsible for the overall technical project administration. To succeed, those developing this type of system require the support of management to get answers to knowledge engineering questions. These answers are not always simple to obtain but can benefit both the design and the operation of complex systems.
Geddes N. D. ( 1989). Understanding human operator's intentions in complex systems. Unpublished doctoral thesis, Georgia Institute of Technology, Atlanta, GA.
Hammer J. M., & Small R. L. ( 1995). An intelligent interface in an associate system. In W. B. Rouse (Ed.), Human/technology interaction in complex systems (Vol. 7, pp. 1-44). Greenwich, CT: JAI Press.