I'm with you on the GPS idea, and it's doable with current technology. Problem is that you can't define where a knee hits. What you COULD do is have the local GPS signal be merged with the video - so you can 'tag' when the knee hits (assuming you can see it) and then use the GPS info to determine the ball location. I use 'GPS' here because that's the commercial tech - you wouldn't use actual satellite info, as it's not detailed enough. You'd have to put sensors on/around the field to get it to be finely tuned enough - only question would then be how does the chip hold up in an actual game.