Can you beat this robot at pool? This sparks something of a “let the wookie win” attitude for us, but we still love to watch the video. This is the PR2 playing pool thanks to the folks over at the Willow Garage. It uses a laser sensor to detect the legs of the pool table, and cameras to find the diamonds and balls at the playing surface rather than using an overhead camera. They cut down on the coding work needed by using FastFiz, an open source Billiards physics library. The final step was building an interface so the robot could use a cue. Check it out after the break (no pun intended).
Continue reading “The PR2 calls the shots”
Sunday we saw robots playing pool and an augmented reality pool game. Today we’ll complete the pool trifecta: virtual pool using a real cue stick and ball in another vintage video from Hack a Day’s secret underground vault. The video is noteworthy for a couple of reasons:
First is the year it was made: 1990. There’s been much buzz lately over real-world gaming interfaces like the Nintendo Wii motion controller or Microsoft’s Project Natal. Here we’re seeing a much simpler but very effective physical interface nearly twenty years prior.
Second: the middle section of the video reveals the trick behind it all, and it turns out to be surprisingly simple. No complex sensors or computer vision algorithms; the ball’s speed and direction are calculated by an 8-bit processor and a clever arrangement of four infrared emitter/detector pairs.
The visuals may be dated, but the interface itself is ingenious and impressive even today, and the approach is easily within reach of the casual garage tinkerer. What could you make of this? Is it just a matter of time before we see a reader’s Mini-Golf Hero III game here?
Enjoy losing at pool? Well the folks at Queen’s University just made it a whole lot easier. The Deep Green robot was created with the purpose of playing a flawless game, allowing it to beat even the most skilled human players. More than a couple of research papers have been written on the project. A ceiling-mounted Canon 350D tracks the position of all of the balls, in addition to another cue-mounted cam (for higher shot accuracy). Using a bunch of calculations, and a computer (probably more advanced than an Arduino), the Deep Green is able to strategize and play. Very well.
On a positive note, another team from Queens is working on a seperate but related project: ARpool (as in augmented reality). It was created to make playing pool easier. The website does not provide much info, but it seems to project different moves onto the pool table, allowing an inexperienced player to tell whether a shot is at all possible.
[Justin] sent in his Digitally Assisted Billiards project. Using a web cam, a computer and a projector, these guys have set up a system that shows you the trajectories of your current shot. It detects the angle of the cue and displays a glowing blue line showing where each ball would go and where the collisions would be. It is a bit slow right now, and made somewhat less accurate by a low resolution web camera. This could be a fantastic teaching tool if it were to get some more polish. The source code is available on the site, so you could try this one out at home.