Can a Robot Beat a Human Player?

[Colin Colenso]

This is like the Deep Blue v Kasparov idea of a computer beating a human at chess.

So, do any of you think a robot, that must play the entire game itself, except for perhaps a human instruction of when it's his shot and which balls he's on, could beat the top pros?

I think it's possible...not likely for a while, but quite possible.

...yes...I'm bored...lol

 

[Solartje]

*steals collin his sixteen 6-packs of beer and goes hide them*

now u behave little boy !!! dont make me steal more of your alcohol

 

[Colin Colenso]

*steals collin his sixteen 6-packs of beer and goes hide them*

now u behave little boy !!! dont make me steal more of your alcohol

I thought you were on you way to have fun with your girlfriend....after your sandwich...lol

 

[Solartje]

that was last night coling remember i live in europe... a good 12h? difference?

and ps didnt get very far. she fell asleep after i called her i'd come over, been standing in the rain for 30minuts at her door, ringing on the door as crazy but she never opened. she was so deep asleep she never heared any bell...had to drive back true the rain on my crappy bike and sleep all alone. well she felt SO bad i should get it double today.

ps if u got so much time... maybe u could use a mathematical formule to count for the amount of squirt u get on a shot, depending on the speed (v (m/s)), power (F (Newton)), tip measurements (mm), weight of cueball (kg), weight of cue (kg), position of touching the cb (X: mm, Y: mm) , place where cb and ob touch ( X:mm, Y: mm), clothresistence (R), temperature of the room (°K), humidity of the room (%), inclination of the cue (°), radius of the tip (mm), amount of chalk on the cb (Kg), gravity depending on the position of the moon (N/m), windspeed in the poolroom (m/s) , density of the chalk used (kg/mm³), resistence of the chalk (R) , packs drunk by the player . i did most of the work already, so all u'll have to do is find the formule. (the easy part )

should give u some work.

ps: dont forget the sexy 3d excel tables.

 

[Guru]

This is like the Deep Blue v Kasparov idea of a computer beating a human at chess.

So, do any of you think a robot, that must play the entire game itself, except for perhaps a human instruction of when it's his shot and which balls he's on, could beat the top pros?

I think it's possible...not likely for a while, but quite possible.

...yes...I'm bored...lol

Have you heard of Deep Green? It's not very advanced yet, but can scratch pretty consistently.

Also, check out the September agenda for the ICGA: Special session on Robotic Pool and Snooker

I don't think a robot can come close to the pool playing equivilant of Big Blue playing chess within our lifetimes. Big Blue had a hard enough time just competing on the computational level when it only had to predict a finite number of legal moves possible. Since the variables of a runout of a rack are nearly infinite, and the dexterity involved hasn't even been mastered yet (though that would be the "easy" part), I'd say it will be a long, long, long time in coming.

 

[Snapshot9]

No way, Jose ...

When we, as humans become equal to God, then yes,
otherwise, no way, Jose. I have been in computers for
over 25 years and developed major systems for corporations
on mainframes and PC driven systems. The infinite number
of possibilities that exist would have to be narrowed to 1 by
reasoning and calculation, where a human can do it merely
by recognition. The mechanics a robot would have to go through
for shooting would never be reached to an acceptable level.

The memory and storage devices required for such a robot would
have a hard time being contained within its shell for Pool. The programming
would be a life long nightmare.

I am still working on Quantum Mechanics, don't put things like this
in my head ... lol

 

[bruin70]

This is like the Deep Blue v Kasparov idea of a computer beating a human at chess.

So, do any of you think a robot, that must play the entire game itself, except for perhaps a human instruction of when it's his shot and which balls he's on, could beat the top pros?

I think it's possible...not likely for a while, but quite possible.

...yes...I'm bored...lol

if you assume the robot will make every shot, calculate every kick, figure out the rails, etc etc.....if a player shoots even 90% he is considered to be playing fantastic pool,,,,yes,,,i think the robot will win most of the time.

or are you asking whether such a robot is possible? i am assuming that such a robot exists. if so,,,,then yes.....but no such robot exists.

 

[Colin Colenso]

Have you heard of Deep Green? It's not very advanced yet, but can scratch pretty consistently.

Also, check out the September agenda for the ICGA: Special session on Robotic Pool and Snooker

I don't think a robot can come close to the pool playing equivilant of Big Blue playing chess within our lifetimes. Big Blue had a hard enough time just competing on the computational level when it only had to predict a finite number of legal moves possible. Since the variables of a runout of a rack are nearly infinite, and the dexterity involved hasn't even been mastered yet (though that would be the "easy" part), I'd say it will be a long, long, long time in coming.

Thanks for the great link Guru. Deep Green is at the beginning stages of what I was thinking.

You all should watch this great video about this pool playing robot!
Watch Here!

If they improve the machine's mechanical and computational abilities it could be a robot with a 40mph break that never misses an on shot and could predict how to play position within a few inches. Enough to beat the best in the world at 9-ball I predict.

In more complex tactical games the task would be much harder.

If more investment goes into pool robots, then I think we'll see them taking on top players in suitably designed competitions.

 

[Guru]

Thanks for the great link Guru. Deep Green is at the beginning stages of what I was thinking.

You all should watch this great video about this pool playing robot!
Watch Here!

If they improve the machine's mechanical and computational abilities it could be a robot with a 40mph break that never misses an on shot and could predict how to play position within a few inches. Enough to beat the best in the world at 9-ball I predict.

In more complex tactical games the task would be much harder.

If more investment goes into pool robots, then I think we'll see them taking on top players in suitably designed competitions.

Excellent, thanks!! My old link to the video was broken, so I linked his homepage. That's what I originally saw on the subject and it really piqued my interest a couple of months ago.

Have you seen this related, yet not robotic project: Automatic Pool Trainer (I want to borrow this thing to help me preserve my relationship with my girlfriend while teaching her to play)

 

[Colin Colenso]

Excellent, thanks!! My old link to the video was broken, so I linked his homepage. That's what I originally saw on the subject and it really piqued my interest a couple of months ago.

Have you seen this related, yet not robotic project: Automatic Pool Trainer (I want to borrow this thing to help me preserve my relationship with my girlfriend while teaching her to play)

My God! How cool is that!

See Video Here!
It's .avi format but should open automatically in recent versions of Windows Media Player.

This kind of table, with the surface basically as a video screen would make the game so much more interesting for beginners.

I'd love to have some advanced versions of these in a pool hall. Showing people the line of the pot etc. It would bring a lot of additional interest to the game.

Thanks for another great link Guru!

 

[LastTwo]

When we, as humans become equal to God, then yes,
otherwise, no way, Jose. I have been in computers for
over 25 years and developed major systems for corporations
on mainframes and PC driven systems. The infinite number
of possibilities that exist would have to be narrowed to 1 by
reasoning and calculation, where a human can do it merely
by recognition. The mechanics a robot would have to go through
for shooting would never be reached to an acceptable level.

The memory and storage devices required for such a robot would
have a hard time being contained within its shell for Pool. The programming
would be a life long nightmare.

I am still working on Quantum Mechanics, don't put things like this
in my head ... lol

Actually Scott, I think you are very wrong. I'm sure you know of pool video games, the best probably being the Virtual Pool series. The toughest opponents in those games play perfectly. You could have a robot take a screenshot of the table and translate it into a game of virtual pool, and work from there.

There are robots in Japan that can already recognize visuals and react to them, so it would be very possible to build a robot version of "deep blue", with hundreds of millions of dollors probably.

 

[5ballcharlie]

My God! How cool is that!

See Video Here!
It's .avi format but should open automatically in recent versions of Windows Media Player.

This kind of table, with the surface basically as a video screen would make the game so much more interesting for beginners.

I'd love to have some advanced versions of these in a pool hall. Showing people the line of the pot etc. It would bring a lot of additional interest to the game.

Thanks for another great link Guru!

Thanks for the video collin it was very interesting. I do not believe that we will ever see anything prevail over the best human players but maybe one day these machines can play and even win. If my memory serves my well. Kasparov lost to deep blue, but he later beat another version of deep blue that was smarter with a better thinking engine. So what I am saying is that even if these things never missed a "OPEN" shot I still believe the best players could beat them with imagination and talent.

 

[alstl]

In virtual pool it has already happened. The challenge is to develop a mechanical device to hook up with the existing software, and unless somebody comes up with a lot of money it won't happen any time soon.

 

[Colin Colenso]

In virtual pool it has already happened. The challenge is to develop a mechanical device to hook up with the existing software, and unless somebody comes up with a lot of money it won't happen any time soon.

The problem with virtual pool is that it doesn't use the exact correct physics such as varying throw with speed and amount of side, cushion reflection angles, swerve etc. It knows it's own formulas and so can make perfect predictions.

That said, a very accurate robot could experiment with all these shots and accurate formulas could be developed.

The process of developing these formulas would produce insights that would keep us physics junkies entertained for endless hours and provide us with an encyclopedia of data to present to annoy and confuse the general pool playing population.

It could play near flawless jump shots, masses and rail kicks if it analyzed the properties of the table beforehand.

Such a robot may be more popular to watch than those Japanese walking robots which develop many millions worth of PR like ASIMO does for Honda.

A robot like that playing pool could generate millions in PR through TV news, TV documentaries and public appearances. It would be worth some investment.

 

[Colin Colenso]

No, because robots cannot 'feel'. They can be programmed with all of the aiming systems in the world, but they just can't feel. Just had to say this

I know you're joking Whitewolf But anyway, just to clarify, a robot can use a systematic mechanism to create the speed of shot required.

Humans cannot do this, except by IJ (Intuitive Judgement / Feel). If we could, we could infact use a system to control our CB position.

 

[Gabber]

I know you're joking Whitewolf But anyway, just to clarify, a robot can use a systematic mechanism to create the speed of shot required.

Humans cannot do this, except by IJ (Intuitive Judgement / Feel). If we could, we could infact use a system to control our CB position.

The robot could have a perfect stroke but I think on how it actually performs depends on the software. Ther are lots of problems but the biggest one is that not all tables/balls are the same. The machine cant react to what is happening on the table so it would have to be able to learn from every shot and make changes according to the conditions.
If you could make such a robot with a perfect stroke that could also play smart, it would be difficult to beat.

Gabber............BTW, I thought your mathematical aiming system post was brilliant. LOL. No really, its a classic. Can you imagine trying to teach that to somebody? "OK, now, just move your bridge 1.7mm to the left!".
It might be a system for the robot ?

 

[Ronoh]

lol

Any machine (when calibratrate) will out perform a human.

Key is, a Human calibratrates a lot faster than any machine can.

A person with minimal intelligence can play pool, and be good at it. To make a machine to do the same thing is sort of redundant, but it's a cool concept.

Bottom line...

The machine will loose, Man will not allow a machine to beat him in a game of pool. Man can always pull the plug

 

[Gerry]

Too many variables in pool to be perfect. No machine will be able to adjust for changing humidity, chalk on the table or cueball causing skids, squirt, slide whatever you want to call it, different weight cueballs, dirty or old balls with imperfections, so there will always be the grey area of imperfection in the playing equiptment to put a wrench in the workings of the machine and thats when the "player" will beat the machine out of all his spare memory cards!!........Gerry

 

[Colin Colenso]

Too many variables in pool to be perfect. No machine will be able to adjust for changing humidity, chalk on the table or cueball causing skids, squirt, slide whatever you want to call it, different weight cueballs, dirty or old balls with imperfections, so there will always be the grey area of imperfection in the playing equiptment to put a wrench in the workings of the machine and thats when the "player" will beat the machine out of all his spare memory cards!!........Gerry

I would say the opposite. That in fact a computerized robot could adjust for all those things much better than a human.

Of course this required lots of testing in advance to derive appropriate formulas for all the variables. And some sensors at the actual table to measure the variables such as humiduty etc. It could measure the friction coefficient of the balls by watching the angle of deflection of one hit and it could also measure the weights of the balls according to tangent line paths. Same with rail properties.

I think the only way humans could challenge advanced robots is by playing more complex strategic games. eg. 8-ball would be more complex than 9-ball and maybe 1-pocket would be more complex than 8-ball.

 

[Colin Colenso]

Oh, in about 200 years from now it will be done, all kidding aside. I read too much science fiction and am way way ahead of you guys . I believe the learning process that will serve the robots is called 'hueristic'. They will be able to learn from their mistakes and build upon their own knowledge. All they will need is the initial human programming spark and they will be off to the races. One day the real hustler will have a robot to do his bidding LOL. Then will will have to have Harrison Ford on hand to perform a robot test hours before the match begins.

It's been a while, but I believe that the super computers which are learning chess as we speak will beat the crap out of humans and they are real real close. As a matter of fact (I have read), the real chess world is indeed afraid that computers will solve the best ways to win and that chess will become dead! The Russians use chess computers to train on all the time now. Just a matter of time.

Well I think it could be done in a few years with a few million invested.

The software is reasonably well developed with virtual pool and virtual snooker. This software can work out strategy and positional play. It would just need some tweaking to make it compliant with a few real world variable.

The robot seen in the video is a pretty simple device, but technology already exists that could make a much more superior cue whacker. It could also be combined with a surface sensitive table that can pinpoint the positions of the balls, saving a lot of complex visual computations. Each ball could have an internal chip so it could be instantly recognized.

Now to put a stick in the hands of a human looking robot like ASIMO, that would take a bit more work. But it would get news coverage worldwide to see some pool challenges between the top robots and top humans.

At the beginning the challenges could be set up to suit the robot's limited capabilities. Gradually real 9-ball and 8-ball could be attempted.

One thing that differs from Chess, is that you cannot use a robot to tell you what shot to play. In Chess, soon anyone with a computer will be able to outplay the world's best, making cheating a potential problem. So chess could in fact struggle in the future.

I think this type of project would be a great investment for a large technology company as part of their marketing. Their robot pool player could travel the technology exhibitions as the prime attraction as ASIMO does.

But right now ASIMO can't run 3 fookin' balls.