Here's a "real" robot playing pool. It looks rather like a giant R2D2, walks around the table (rolls, really), has arms and uses a real pool cue and bridge (an open bridge rather than a closed one). I don't know if it can chalk its own cue, but doubt it. The other pool-playing "robots" I've seen look like travel lifts, don't use a cue, etc.
http://www.youtube.com/watch?v=HMx1xW2E4Gg
I looked at the shot-planning part of the code for this thing. In effect, it uses the ghost ball method for aiming: it calculates the position at which the center of the cue ball and the center of the target object ball would be in line with the target pocket. The comment in the code at the start of the calculation makes that explicit:
// compute cue ball target position in the table frame
// this is the pose of the cue ball center at instant of contact with target ball.
// on contact, centers of the cue ball and the target ball should align with the pocket.
(This portion of the code is in C++; much of the rest of it is in Python.)
This is obviously the simplest way to aim for a computer. The advantage it has over a human using the ghost ball method is that it can calculate exactly where the cue ball must go.
It would be quite interesting to see an algorithm, with pseudo-code to implement it, that accomplished the same result using any of the pivoting systems. Maybe then we would finally begin to understand the geometry that makes them effective.
http://www.youtube.com/watch?v=HMx1xW2E4Gg
I looked at the shot-planning part of the code for this thing. In effect, it uses the ghost ball method for aiming: it calculates the position at which the center of the cue ball and the center of the target object ball would be in line with the target pocket. The comment in the code at the start of the calculation makes that explicit:
// compute cue ball target position in the table frame
// this is the pose of the cue ball center at instant of contact with target ball.
// on contact, centers of the cue ball and the target ball should align with the pocket.
(This portion of the code is in C++; much of the rest of it is in Python.)
This is obviously the simplest way to aim for a computer. The advantage it has over a human using the ghost ball method is that it can calculate exactly where the cue ball must go.
It would be quite interesting to see an algorithm, with pseudo-code to implement it, that accomplished the same result using any of the pivoting systems. Maybe then we would finally begin to understand the geometry that makes them effective.
