IIRC the sensor uses optical flow and hence doesn't measure speed "directly" -- in spite of the fact that the article states so at its very beginning. Anyway, this technique is kind of promising and I'm curious to see the future results.