Justin Bergman is on a mission
- Thread starter JAM
- Start date
If there was incentive to do it an AI pool player could exist in less than five years. Ever growing lack of incentive might keep robot pool players at bay or some of the whiz kids at MIT or a similar place may do it for a class project. It may have already been done.
We will see more advances in technology in the next five to ten years than we have seen in the last twenty. I think of my brother sometimes, a very high level genius who died in the early seventies. He never saw the completion of the interstate system, a microwave, a home computer, ... Somebody else worked on a high level military computer. If it was down thirty minutes the Pentagon called to find out why. It had logic gates that were about three by five inches that you could flip open to fix! The ten meg hard drives were the size of a big file cabinet and the individual disks were the size of 78 LP's! Roughly twelve inches. Won't be long before almost nobody knows what a hard drive is. I need to read some DVD's right now, me or nobody in my family has equipment that can read one! Thirty years ago I could have taught how to work on computers and networks, at any level. Now I am happy if I can just turn this damned thing on!
About Jennie, anybody that wants in a grammar or typing battle with somebody that used to transcribe reams of documents for the United Nations, often from recordings of broken English to proper spelling and grammar, and on a rush basis, ain't going to get my backing! She seems to be semi retired now but I suspect she can still type over three hundred words a minute, no flaws!
From what I read from AI, I can still spell and have better grammar than most programs. Some of the stuff I wrote for bids going to NASA would probably look like AI to a lot of people today. A lot of the reason I am so verbose, the request for bids specified nobody would get any follow-up contacts, no questions, no requests for clarifications, you got it right the first time, or it got round filed. To make things tougher, the requests for bids had errors and you couldn't ask for clarification concerning those errors either!
Hu
I've said pool is the quantum mechanics of tic-tac-toe. Robots can already replicate complex mobility and balance. The issues in pool robots would probably be mostly targeting.
I think the key is keeping them off drugs, so many human players were destroyed by drugs
g0ds gReeN
Active member
Circuit to edge? Ghost transistor? Ghost IN the ball?The real question is what aiming system would the robots use?
Im scurred
Without an extremely complex model backing up the laser sight and straight stroke, the robot would miss all the time.
It doesn't just aim, it also has to model squirt correction, swerve correction, and throw correction. It would need to adjust for cloth and balls that play differently. A post-contact vector for position, velocity, and spin would need to be estimated and then highly nonlinear rail behavior would need to be modeled and, like the cloth and ball information, learned on the fly.
So we are talking about
- physical world model
- initial calibration of equipment
- continuous on the fly learning of conditions
- full table perception via computer vision
- probabilistic short and long term shot planning - evaluating 3-7 shots ahead, position zones, and strategies, not just one shot
Oh yeah, and actually creating a robot that moves around the table and is able to chalk the cue, take a stance, and deliver the stroke. Chalking in particular seems like a nontrivial skill. Contact needs to be made to millimeter-level precision.
The best practical robot would probably simplify as much as possible, play close to center ball, use natural angles, do all sorts of things to minimize the risk of bad things happening. We won't be seeing one any time soon though.
CPG of course. They can throw in exogeometry - that would be Nasa going it can't be this easy.The real question is what aiming system would the robots use?
Isolation from their creators would be the only solution and then you'd have Irobot.
Well we'd both have to live 20 years for me to collect. So the only way I'd make the wager is if it was to only collect after the 20 years regardless as to whether the robot does it within the time frame and we both have to live the 20 years, but I'll happily make the wager under those conditions.
Actually the most difficult issue in pool robots is the difference in conditions, the variability of cushions, etc. That's one of the reasons that I clarified that by autonomous robot, I meant able to walk into a random pool hall with a random set of balls that aren't brand new or freshly polished and autonomously choose shots. No special lighting conditions or markers allowed.
I could make a robot right now that would win if I could design it as a part of the table and use special markers and lighting.
I wonder if they could have simple compensation - like humes. Impulse response a table in a few minutes and then kick ass. With actual AI they could hustle too - make truly impossible games.
