Developing a Knowledge Engineering Capability in the TRW Defense Systems Group

The TRW Dcfensc Systems GI oup develops large mailmachine net,works that, solve problems for govrrmnent agencies. IJntil a few years ago these networks were tither tightly-coupled humans loosely supported by machineslike 0111 ballistic missile system engineering organization, which provides technical advice to the Air Force, or tightlycouplrd machines loosely controlled by human~likc the ground station fox the NASA Tracking and Data R.elay Satellite Syst,em. Because we have been producing firstof-a kind syst,ems like these since the early 195Os, we consider ourselves leaders in the social art of assembling effective teams of diverse expcrt,s, and in the engineering art, of conceiving and developing networks of interacting machines But in t,he mid-1970s WC began building systems in which humans and machines must be tightly coupled to each other-syst,ems like the Sensor Data Fusion Ccntcr (Figure I). Then WC found that our well-worked system development techniques did not complet,ely apply, and that our system engineering handbook needed a new chaptIer on comnnmication bct,ween people and machines We’re still writing that, chapt,er, and it won’t, be finished until we can add some not-yet understood cognitive psychology and some not-yet fully developed artificial intelligence t,echniques Nevertheless, we have learned some lessons worth passing along.