Lego Robot Solves Any Rubik's Cube In 12 Seconds

kkleiner writes "Cube Stormer is the latest creation from Mike Dobson, aka Robotics Solutions, and not only is it made entirely out of Legos, it can solve any 3x3 Rubik's cube in less than twelve seconds. Often it can finish in less than five! This thing looks bad-ass and is incredible to watch."

That's fast

[TheVoxyn]

Think 99% of the world population can't do that :P

How they are doing it?

[JoshuaZ]

Unfortunately, the linked to post and video doesn't give much details. Naively, I expect that the computer program is first figuring out very quickly what the series of movements to solve the cube and then implementing those. There are around 4 * 10^19 possible configurations for a Rubik's cube, but the group theory allows one to work out what steps to take without having to do very exhaustive searches since the Rubik's group is very well-behaved. However, this assumes one is in an actually solvable configuration. I'd be curious to find out if they've debugged the device well enough to make sure it doesn't hang or get in some infinite loop if one gives it an unsolvable cube (not all possible permutations of squares are solvable. Most trivially, edges need to stay on edges, corners on corners and centers on centers. But some configurations are still not solvable. For example, if one swaps two center stickers it isn't hard to see that that lays outside the Rubik's group of reachable permutations).

Re:How they are doing it?

[MBCook]

It looks pretty simple to me. You put it in and it snaps shots of the 6 sides of the cube. Those are interpreted by the computer which probably uses a standard solving algorithm. The solution is translated into movements for the robot, and off it goes.

My guess would be if it was impossible to solve, it wouldn't start doing anything, the software would complain. No Rubik's cube is impossible to solve without physically messing with the cube (as you pointed out, swapping stickers for example). If you start with a solved cube, no amount of twisting can make an impossible cube.

The video is quite impressive, far better than most lego solving robots. I'd love to see this thing solve a bigger cube.

Or how about a feeding device? You put 10 cubes in, each is automatically placed in, solved, and popped out.

Simplified hardware

[LoudMusic]

This guy did it a while back with considerably less hardware, though it takes his rig a bit more time to get the puzzle done ;)

http://www.youtube.com/watch?v=htnL1KTpaY8 [youtube.com]

I don't get it

[computerchimp]

I can't understand why this is a "Lego" robot.
The pads are Lego the rest of the Lego is total cheap fluff. If I stick a few pieces of Lego on my car does that mean I drive a Lego car?
Maybe it is some cheap promo.

What am I missing here?

CC

Re:Obligatory

[mgblst]

The main problem is that the US calls it one way, the rest of the world another way.

Just as with metric, math (vs maths), the US has to do it differently.

Re:That's fast

[turing_m]

Are you sure? Just about every Asian I know can do it that quickly, and they make up about 30% of the world's population.

Interesting comment. East Asians have a higher visuospatial IQ. It would make sense that solving a Rubik's cube would play to their strengths - it's pure visuospatial ability. Your anecdote rings true with me - I remember being amazed at how quickly a group of average Japanese students could play Tetris on the Gameboy. They were able to play it indefinitely at the fastest level. The ability of their brains to rapidly process that sort of information - this block rotated this way, will fit there - was much more efficient than my own, even with practice. And it's not that I'm uncoordinated or bad at video games in general - far from it. In FPS or RTS games I'd easily be well within the top 10%. But with a game like tetris, I was not able to use reflexes, hand-eye coordination or intelligence in a strategic sense to make up for my slower visuospatial processing.

This sort of experience makes me suspect that there is a difference in mental "modules" between Europeans and East Asians, and if you don't have the right "module", it's like trying to play a modern FPS without a 3d graphics card. You might be able to do the same task in a "software emulation" type mode, but it will be at reduced speed. Or maybe if it is too hard, your brain just can't do it.

Re:This Video is FAKE

[bondsbw]

The next to last solve (at 41s) takes 21 moves*, and is the only cube claimed to be random... thus, I don't see any dishonesty. It takes around 1.9 seconds to analyze, about 0.4 seconds to reset/process, and the remaining 8+ seconds to solve. Therefore, it makes on average between 2 and 3 turns per second.

Humans do not include inspection time in the speed calculation (at least, that's the case in the accompanying video of the world record). An apples-to-apples comparison, therefore, would be the human time at 7 seconds and the robot at a little over 8. I couldn't follow the world-record video, but I think I saw at least one mistake (a move followed by the opposite move) and a little hesitation. So, you're probably correct in the 3-5 moves per second for humans.

*21 includes twice that the computer simultaneously moves two faces, each counted as two separate moves. 180 degree moves are counted once.

Re:How they are doing it?

[Paiev]

I'm a competitive cuber/scrabble player, you insensitive clod!

(No, seriously, I am. Scrabble has a lot of depth to it when played on a higher level that you are completely ignoring. Don't be so quick to dismiss something just because you don't know much about it.)

Slower manipulation, faster thinking

[The_Duck271]

Mostly this is just a demonstration of how a computer can, from the initial scrambled state, immediately see clear through to a solution in a relatively short path, whereas humans can't visualize a whole solve instantly, and so they take it in steps, at a significant cost to solution length. Comparing the two videos you can see that the human is much faster than the robot at making sequences of turns, but must make many more moves than the robot.

Re:Stickers

[JosKarith]

When I got stuck on the Rubiks cube, I worked out how to disassemble and reassemble the whole cube in the correct order. Peeling off the stickers never even occured to me.