Louise Marston - 16.01.2012
Kinect does very sophisticated visual analysis, interpreting a picture of your body to work out where your hands, feet and head are, and uses this to control X-box games. It knows if you move towards and away from it, as well as side to side, and can distinguish more than one player.
You would expect this sort of 'high tech' to come out of advanced computing laboratories at universities like Cambridge and MIT. And you'd be right. But that's not the whole story.
Microsoft has a large and influential research function, closely connected to universities, which publishes a great deal of original research. Speaking at the UK-IRC Innovation Summit in November, Andrew Fitzgibbon from Microsoft Research in Cambridge described the role they played in developing Kinect. The Kinect business group in Seattle had seen a paper published by the Cambridge team, and asked them to help build a system to track the human body from images. The research team advised them, based on years of study on computer vision, that they were asking for the impossible. It couldn't be done fast enough, it required too many calculations to be done in real time.
No, no, no, the Kinect product team said - we've already built the thing, and it's working. We just need your help to improve it. Later, a film computer graphics company was brought in to the collaboration, creating hundreds of human pose images to train the system to recognise every possible configuration of the human body.
The story of Kinect brings together two common stories about technology development: that it comes from research and academic study, and propagates out; and that it's all about studying users, and designing what will delight them. The example of Kinect illustrates that you need both (and usually many more components as well). You need the research, the detailed study of what works and how. But you also need to connect that to those who can see the needs and opportunities, and just try something out.
When we think of 'technology' as being pushed out of laboratories, or pulled along by the voracious suck of the market, we are missing the biggest piece of the puzzle: the connections and collaborations that bring together existing work and knowledge in new and surprising combinations. All of those are built on human relationships, and we are much further away from working out how to influence these than we are in the study of university knowledge transfer or venture capital.