You’ve heard of Deep Blue, the IBM computer that bested Gary Kasparov in a chess match a decade ago. Now, there is Deep Green, a robot that plays pool. And by the looks of the demo video above, it can’t lose.
As Delicious founder, and now-Googler, Joshua Schachter points out, it is “only a matter of time before one of these kills a person.”
Deep Green is a project out of the robotics and computer Vision lab in Queens University. They are also working on an “augmented reality” version of pool that lines up your shot with lasers. This is also known as cheating.
It’s kind of hard to be a pool hustler when you are on a robotic chassis and don’t have a face. I’ts pretty easy to spot and avoid playing them.
Sadly my first thought when the AR came on was that in 5 years someone’ll build the tech into sunglasses and then everyone with enough $ could become a pool shark.
Pankaj Advani watch out, here comes Deep Green!
oooo… i want one! does it know how to do “english”?
I’ll be impressed when it can assess all the options on the table after a break and line up a whole series of shots in which each one leaves it in position for the next. Making one shot is not winning a game of pool.
I can wait for the Iphone app… haha
Very cool.
Although, like any other AI, its subject to exact precision… and in this case thinks geometrically and not 3 dimensionally, inevitably misunderstanding the nature of the table felt and its impact in Pool. A touch of Chaos theory and how as the ball travels, it racks up micro changes in the felt culminating in different results then calculated.
Still, very cool project to be working on!
I’ve had the idea of a pool playing robot for a while now and this was a pleasant surprise story to read.
The issue that you describe can be alleviated by letting the robot “calibrate” it’s algorithms by playing simple “lag” shots at various points on the table as well as playing a preset number of 2-rail, 3-rail and 4-rail shots.
No tuning will ever be perfect, but this would be a good starting point to estimate the hardness of the rails, table roll and table felt issues.
Rack um’ Deep Green, I will be back in an hour or so…
With Deep Green, I’m pretty sure you’d be the one doing all the racking
.
And we move another step closer to creating our new robot overlords.
It will be cool to see the bologna between the bread if you catch my cupcakes.
No I don’t catch your cupcakes but I do hear the marbles rattling in you head though…
Those are not marbles… Loose screws.
Cool….but
when are we going to get a robot that can shoot hoops or something?!? and what’s the future of competitive sports.
AWESOME!!! the augmented reality could be VERY popular… in bars.
I don’t want to brag, but I’d *destroy* Deep Green in a game… okay, maybe I do want to brag. Awesome technology though.
Agreed.
Having played Professional Pool from 96 – 2002 I would really like to see how well I would do against a computer program.
After seeing the robot make a series of uber noob shots I don’t think it would fair so well against a skilled player who can use strategy and safety shots to manipulate the table to his advantage.
Also I don’t have the patience to wait 5 minutes while it racks a 9 ball rack.
its pretty obvious that in a decade or two, human evolution is going to move away from the darwinian and into the digital.
i cannot imagine hwo people will be able to resist bio mechanical computational augmentation. i expect in 50 years the human brain will be similar to the matrix in that, people will have implants in the brains, to which they will be able to download software, efectively programming themselves in a second to learn, anything, everything.
instead of imaging a reason for it to happen, im struggling to imagine why it wont.
Then there’s the “real” Deep Green, the widely popular chess application for iPhone. It first came out for the Apple Newton (hence the “Green”) the year after Kasparov was defeated by Deep Blue. http://cocoastu...ducts/deepgreen.
Very similar to my 4th year design project at Waterloo – two years ago!
We didn’t have a robot – we simply projected lines on to the table so you could “cheat”. It took into account spin, multi-ball collisions, level of force, skill level and a bit of post-shot positioning.
Lots of fun.
are you john connor, son of sarah connor?
only nerds would ever think of doing this to pool.
Dammit, Queens University Canada not Queens University Belfast. Why can’t my alma mater ever do anything as cool as this.
I’m working on a robot that can automate the decsision process of the fed chairman.It will be called DEEP SHIT.