/sci/ - Science & Math

>>4239350
Looks pretty optimal to me.

OP, you just have a little bit of machine vision to read the cube's initial state, an internal algorithm for finding the quickest solve, and some electric motors that rotate arms that are gripping onto the cube faces. And a nifty elevator mechanism.

Yes, it is neat.

isn't the mid column of the front face blue?

Computers can use algorithms that people can't
Learn2grouptheory

Totally possible, but it looks fake. I doubt anyone actually built this thing.

It's built with NXTs. I had to use those things for a bunch of shit in one of my classes.

display image doesn't match puzzle at any point
cannot solve with one camera view and no automatic rotation
the big issue is the software, which is ignored -- the picture shows the rotation of the sides,
which would be easier

>>4240209
>display image doesn't match puzzle at any point
blind? check it out once more, display is showing the first seconds, then a simulation.

>cannot solve with one camera view and no automatic rotation
blind? when the camera captures, the cube IS being rotated along two axes to get the initial state.

better be trolling

>>4240570
OK, this is interesting: you're saying the screen shows the initial state, then goes into a (pointless) simulation graphic?
So, it's showing a camera view, then changes to a non-camera view, even though it seems to be the only camera and processor visible?

When I wrote about not rotating, I was writing about the camera -- the camera needs to see more than one side to solve the puzzle.
(it could automatically rotate if it is using just one camera, but it does not).
That is, pieces are being rotated, but not the whole cube at each step.
it is possible, but would have been much harder, to solve the puzzle by photographing all sides in the first case, but that doesn't happen, either.

It would be fun to give it an unsolvable cube.

>>4240604
>OK, this is interesting: you're saying the screen shows the initial state, then goes into a (pointless) simulation graphic?

>you're saying

I'm not that anon, but "he's" not "saying" anything-- that's what the fucking gif shows.

I'm not sure why anyone would be very surprised by this.

I mean, fuck, we have video games you can dance in front of now.

>implying rubiks cubes don't get stuck all the time

I'm guessing some piece would fly off

This is old news. How have you people not seen this before. And it is real

>>4240632
people playing these actually oil the pieces or some shit to reduce friction

I'd say its real
You have to slow it down so you can see what its doing. Its just running through it so fast its hard to catch everything its doing.

We've had computers that could beat the best chess players on earth for decades. How is this shocking?

what is the surprise here?
old news.
Rubik plus Lego NXT and android phone.

>>4240604
You don't need to watch constantly for all the faces all the time.
You just need the initial state of each face. Look for best algorithm. And proceed.
And that is what they did.

Give a 4x4x4 and make it recognize/solve parities, then maybe I'll be impressed.

http://www.youtube.com/watch?v=_d0LfkIut2M&feature=related

>>4240696

Give it*

>>4240623
The images shown do not match the cube below:
I wrote that it mismatched, therefore wasn't suggesting processing or using the camera as sensor.
He wrote that the display showed the initial state, then went into a simulation graphic (deliberately).

I think that's nonsense: imagining a pointless simulation graphic rather than just assuming the whole gif is a fake.

>>4240661
Not the least bit shocking, nor that the puzzle can be solved:
the question is whether the images shown are of a device doing that.
But, lest you take too much credit: chess devices, until recently, have not actually moved the objects, and chess doesn't require the same amount of visual interpretation (or none).

>>4240653
Seen on 4chan, or seen in genuine press coverage or school news, in any form other than a gif?

>>4240689
"And that is what they did."
Don't know why you assume that.

"You just need the initial state of each face. Look for best algorithm. And proceed."

In the video, we see the cube put into the device and the initial state of a single side appear on screen. None of the other sides, which you just said is necessary.

It is certainly possible their algorithm doesn't checks just one side and solves that one first, but it would still need to see the other sides to find which parts to move into place.

>>4240703
OK; I recant.
The youtube video is far more clear about how the device checks initial state; it plainly rotates the cube to check sides.
In the gif, that is happening in frames 40-55, but it is not anywhere near as clear.

OP, you are right. That machine does not exist. We all know humans are the only ones who can do it, because they are the only ones with souls.

>>4240863
This. There's simply no way for an advanced computer, especially a tiny one inside a cell phone to do that.

It takes human skill to do this. Trust me, I've tried it. These puzzles are TOUGH. Even I couldn't ever solve one. There's no way a computer could either.