Robotic Pool Player?

On a similar note there is professor at Queens University in Kingston developing a robot to play pool. His ultimate goal is to have it play at a professional level, kind of like Deep Blue, only for pool.

Unfortunately when I last saw a feature on this they were having some tough times. Especially with the break. Many of their problems are a result of not actually knowing anything about pool, outside of the basic physics. Figures.
 
Cameron Smith said:
... Unfortunately when I last saw a feature on this they were having some tough times. Especially with the break. Many of their problems are a result of not actually knowing anything about pool, outside of the basic physics. Figures.
Click to expand...
But the break shot (straight at the front ball) is pretty nearly the simplest shot on the table. If that can't get that right, they need some serious help with the basic physics.
 
Bob Jewett said:
But the break shot (straight at the front ball) is pretty nearly the simplest shot on the table. If that can't get that right, they need some serious help with the basic physics.
Click to expand...

The issue had to do with a lack of power in the break. It was pretty bad too as they were not able break very well on a bar box :eek:.

The problem I noticed was that the mechanism didn't have much of a follow through at all. From start to finish I would say the stroke was about three to four inches. Tough to create much power that way, machine or not.
 
They should hire Allen Hopkins as technical advisor.

Cameron Smith said:
The problem I noticed was that the mechanism didn't have much of a follow through at all. From start to finish I would say the stroke was about three to four inches. Tough to create much power that way, machine or not.
Click to expand...
 
I'm not sure about the group of scientists ... a grad student for sure, and he likely had some help with various elements of the project. He has moved on to robotic snakes ( http://www.inference.phy.cam.ac.uk/rpa23/snake.html ), so I think this 1999 pool robot may be collecting dust now. It was likely just a device to use in his studies on inference, but certainly a fun one ! Anyway, this project makes some some of mine look very professional. I loved the saw horse frame !

BTW, I think you can develop significant power in a few inches. Hopefully some day I will be able to prove that.

Thanks for the link Linda, I for one am very interested in these devices.

Dave
 
Ya gotta admire their enthusiasm

There's just something cool about a bunch of inventors getting together to do something like this.

However, even if they could make a machine that read the location of the balls and execute the shots perfectly, the math required to allow a machine to make the right or best choices on a real table against a real player seems to me would be absolutely staggering compared to a chess game. At least in chess, you start with 64 spaces and 32 pieces, basically a finite number or definable number. In pool the mathematical variables are infinite. It would be pretty interesting to see how far they got with this part of it! :)
 
Da Poet said:
... However, even if they could make a machine that read the location of the balls and execute the shots perfectly, the math required to allow a machine to make the right or best choices on a real table against a real player seems to me would be absolutely staggering compared to a chess game. At least in chess, you start with 64 spaces and 32 pieces, basically a finite number or definable number. In pool the mathematical variables are infinite. It would be pretty interesting to see how far they got with this part of it! :)
Click to expand...
I think the AI player in Virtual Pool does a reasonable job. It doesn't have to find the best shot, just a series of shots it can make, and the best level of computer player is an accurate enough shot that it doesn't need perfect position. Mosconi it's not.
 
Cameron Smith said:
The issue had to do with a lack of power in the break. It was pretty bad too as they were not able break very well on a bar box :eek:.

The problem I noticed was that the mechanism didn't have much of a follow through at all. From start to finish I would say the stroke was about three to four inches. Tough to create much power that way, machine or not.
Click to expand...
I have heard of some high-speed film (actual film, not video) that was shot of a stick hitting a ball in the 1950s. Considering the group that was involved and some of the results that have been muddily reported, I suspect they used an explosive charge that used a trigger from the film camera. Should be possible to get up to break speed in an inch or so of movement. If you don't want to shoot the people behind you, it'l be harder.
 
rackmsuckr said:
I came across this article. It may be old, but I thought it was interesting that a group of scientists would try to mimic all the mechanics of making a ball. Turned out to be tougher than they thought.

http://www.inference.phy.cam.ac.uk/rpa23/billiards.html
Click to expand...
Here's a Youtube video of a similar machine actually pocketing balls. Well, it's no Earl or Efren, and that's not a Diamond Pro with Simonis, but it does seem to work.

http://www.youtube.com/watch?v=J70p-dCfi_s
 
Masayoshi said:
whats it do when the cue ball is on the rail?:confused:
Click to expand...
It jacks up as high as it can and applies an extreme amount of left hand or right hand WD40.

Boro Nut
 
Bob Jewett said:
Here's a Youtube video of a similar machine actually pocketing balls. Well, it's no Earl or Efren, and that's not a Diamond Pro with Simonis, but it does seem to work.

http://www.youtube.com/watch?v=J70p-dCfi_s
Click to expand...
Given that it's just physics I think it's safe to say the technology to build the perfect pool playing robot already exists. I can't imagine why anyone with the necessary knowledge or funding (eg ABB, Mitsubishi etc) would bother, unless it was to demonstrate the dexterity and precision of their latest commercial assembly robots.

They already operate well beyond human capabilities with regard to precision and repeatability. Any error on a pool table would be a measure of the programmer's inability to define the operating parameters, not the robot's ability to perform the task faultlessly.

It would also be relatively easy to build in a feedback loop to teach the robot to temper it's playing based on the outcome of all it's previous shots. It would be up to speed with the table conditions while you were still scratching your head trying figure out what went wrong.

Boro Nut
 
Bob Jewett said:
Here's a Youtube video of a similar machine actually pocketing balls. Well, it's no Earl or Efren, and that's not a Diamond Pro with Simonis, but it does seem to work.

http://www.youtube.com/watch?v=J70p-dCfi_s
Click to expand...

It does work, although the speed is always the same. I wonder if they are going to vary the force of the hit. I would assume it to be easy to do, except for the calculations needed....do I want to break out the cluster up table, where will the balls go after I do, or do I just want to draw a little bit for position on this ball instead, etc.

It looks like this is merely an extension of a person behind it controlling it, sort of a mechanical cue, and not really a thinking robot. I would think that would be very tough to recreate a full table run, especially if there are clusters or even if you wanted to play safe. And how would it make that decision?

Also, does anyone else think the balls roll funny after the hit? They just stop like they have lead weights in them, even though they are hit firmly.
 
You must log in or register to reply here.
Back
Top