A Nice Gesture by Jeroen Arendsen

A Nice Gesture by Jeroen Arendsen

A mix of posts on gesture, HCI, perception or robots and some personal information.

A Nice Gesture by Jeroen Arendsen RSS Feed
 
 
 
 

Double Movement Detection

As a young researcher I see opportunities for grand research proposals everywhere. And so it was when my friend Edwin told me about the rugby drama he had experienced in an Irish pub in the final day of the recent Six Nations.

Ireland beat Italy with a monsterscore, and they were about to win the tournament. But the French scored in overtime to beat Scotland, just edging themselves in front of the Irish and claiming Six Nations victory in 2007.

That in itself is tragedy enough for Irish fans to start some serious forgetting/drinking. But it gets better: The final French try was an almost impossible call. The ball was buried in a pile of players, and even the video ref (who was Irish!) could not make the call. It was granted nevertheless.

Now to the point: in extra time at Ireland-Italy an Italian player scored a try that was contestable. The Irish in their pub were convinced that it was a case of double movement (or even triple movement). This requires some explanation.

BBC Sport: “Often when a player has been tackled close to the try line, they will often attempt to make another movement to ground the ball for the try. However, if they have been tackled, the referee will not award the try because it is seen as a double movement [if] the ball and the player have been grounded before the second movement for the try. However, if the player is in the process of being tackled and the ball has not been grounded before the try line, then they can make a second movement for the score.

A nice case of a visual perception task. The referee has to see that a player makes two separate movements and not one big one. A simple task? I dare to disagree. What exactly makes us see a movement boundary? Do we all agree on it, or did the Irish see double movement (around 4:30 in the movie) and the Italians only one?

Rubin and Richards (1985) did some nice work on Boundaries of Visual Motion that I think definitely applies. But there has not been much work on actually implementing their ideas into algorithms that I know of.

Could we build a computer to automatically segment movement? Can a robot referee call double movement? I am thinking of the automated line judgements in tennis. It would not be the first time people relied on machines rather than their own perception. When we think of time measurements for racing machines are possibly trusted even better than men. In this case I think more research is necessary.

If you agree and have R&D money to spend, give me a call at +31 15 2783908. I’ll be looking for a nice chunk of basic research work on visual perception somewhere next year I presume. Full Game Report at Irish Rugby:

Leave a Reply

Categories

RSS My Del.icio.us Links

Archives

RSS Books in My Library