One of the early human modeling project we worked on was a marksmanship trainer for the Navy. We converted our Jack model into a real-time system. We pre-scripted movements, combined them into a posture and motion transition graph, and drove the system from a server elsewhere via standard distributed interactive simulation packets called PDUs. Jack was pretty good at running, walking, crawling, shooting, standing, kneeling, and dying. But we had mostly engineering students in those days, and they weren't the best artists nor animators. Consequently our Navy sponsor sought a commercial contract to improve both the visual appearance and motion veracity of the interactive human model. The sponsor awarded this contract to Marc R.'s company Boston Dynamics, Inc. (BDI). BDI had seen the Jack version of the system and consequently built a better soldier model and gave it motion captured behaviors. The movements were naturally better, since they came from a soldier subject. The BDI system became known as DI-Guy and went on to dominate the human model component of the distributed interactive simulation world. We finished our relationship with the sponsor now that a commercial product served its needs
Some years later, in talking to Marc R., I congratulated him on his sensible approach to DI-Guy's motion simulation through human subject motion capture. He told me that he contracted that job out to a midwestern motion capture company, and that the data was collected and processed by one Brian S. It turns out that Brian had actually worked in my lab as an undergraduate and had had some experience with Ascension Technologies motion capture equipment. So one of my own students did the work that inadvertantly took us out of the real-time distributed interactive simulation world and promoted our replacement, DI-Guy .