Slashdot Mirror

← Back to Stories (view on slashdot.org)

Android Phone Solves Rubik's Cube In 12.5 Seconds

Posted by timothy on from the varies-with-the-cube dept.

DeviceGuru writes "A Lego Mindstorms robotics kit controlled by an HTC Nexus One smartphone successfully untangled a Rubik's Cube puzzle in 12.5 seconds at this weeks ARM technical conference in Silicon Valley. The current 3x3x3 cube-solvers's 15-second average represents a substantial improvement over the 25-second solutions of an earlier version, which was powered by a circa-2006 Nokia N95 smartphone, thanks to a faster (1GHz) CPU, more RAM, and revamped cube-solving algorithms. ARM Engineer David Gilday, who created the robotic cube-solver, claims the current version's algorithms can handle cube complexities up to 100x100x100, assuming he build the mechanics. In terms of racing humans, Gilday says the Lego robotics kits can only manage around 1.5 moves per second, whereas human players can make between 5 and 6 moves per second, amazingly enough." Update: 11/12 03:45 GMT by T : Apologies to creator David Gilday, whose name was earlier misspelled.

76 comments

  1. so... by gEvil+(beta) · · Score: 4, Funny

    In terms of racing humans, Gilder says the Lego robotics kits can only manage around 1.5 moves per second, whereas human players can make between 5 and 6 moves per second, amazingly enough.

    So, if we could build an ARM-powered human, it could solve the Rubik's Cube in 2-3 seconds?

    --
    This guy's the limit!
    1. Re:so... by Anonymous Coward · · Score: 5, Informative

      most humans have arms

    2. Re:so... by MichaelKristopeit161 · · Score: 0, Interesting
      the problem is the qualifier "CAN make"... the precondition being only certain subsets of 5 and 6 moves can be made in 1 second. the moves necessary to solve the cube will not always fall within those subsets.

      i can solve any cube in under 3 minutes and that's fast enough for me.

    3. Re:so... by Monkeedude1212 · · Score: 4, Funny

      +1 Informative.

      I did not know that.

    4. Re:so... by Kjella · · Score: 1

      So, if we could build an ARM-powered human, it could solve the Rubik's Cube in 2-3 seconds?

      Yes, but if you wanted to you could probably build a high speed, high precision tool to solve it in way less than a second. It's Lego after all, it's cool but it's not exactly built for performance.

      --
      Live today, because you never know what tomorrow brings
    5. Re:so... by MichaelKristopeit172 · · Score: 2, Funny

      how long does it take to shove up your ass?

      It depends if I do it 'corner first' or if I try to make it fit 'flat side up'. Either way I can top the 5 - 6 moves per second.

      I am PATHETIC.

    6. Re:so... by MichaelKristopeit161 · · Score: 1, Informative

      the precision is limited by the joints in the cube... if you are eliminating the physical act of manipulating the cube, then yes, you could solve in under 1 seconds, but i highly doubt a machine could physically solve a randomized cube in under 1 second including analyzing and computing time. a custom cube with custom bearings and springs might help to get the time lower, but then you're not really solving a rubik's cube.

    7. Re:so... by blair1q · · Score: 3, Funny

      I would have thought it was obvious:

      http://i.dailymail.co.uk/i/pix/2008/11/08/article-0-0266436E000005DC-774_468x692.jpg

    8. Re:so... by Anonymous Coward · · Score: 0

      Meh. Been there, done that: http://www.youtube.com/watch?v=PMWWN07PShA

    9. Re:so... by blair1q · · Score: 1

      Put the "robot" in the cube's hub. Use fast motors. Titanium or carbon fiber to get strength and low weight from the faces and axles.

      1 second might be doable.

    10. Re:so... by DeathElk · · Score: 1

      Wow, that was trippy...

    11. Re:so... by MichaelKristopeit161 · · Score: 0
      you're no longer solving a rubik's cube... you're just solving something that looks like a rubik's cube.

      why not replace the stickers with small color displays that adjust themselves in 5ms?

    12. Re:so... by NatasRevol · · Score: 1

      This would be more helpful.

      http://www.worldofstock.com/slides/PCU2137.jpg

      --
      There are two types of people in the world: Those who crave closure
    13. Re:so... by blair1q · · Score: 1

      Maybe not, if this is what it serves up:

      http://www.worldofstock.com/images/hotlink.jpg

    14. Re:so... by Anonymous Coward · · Score: 0

      Humans also have 1 ball, 1 breast, and half a dick.

      On average.

    15. Re:so... by Arthur+Grumbine · · Score: 1

      I like Michael Kristopeit because he keeps me aware of the current /. UID count. All the better for me to try to snag 1999991, 1999999, 2000000, or 20000002 when the opportunity arises in a couple months. He's also, through his many accounts, probably the most powerful moderator on /.

      --
      Now that I think about it, I'm pretty sure everything I just said is completely wrong.
    16. Re:so... by eltaco · · Score: 1

      so copy the link and open it up in a new window to get rid of the referrer.

      --
      It's not about fate, it's about character.
      there be no shelter here, the frontline is everywhere!
    17. Re:so... by MichaelKristopeit172 · · Score: 0

      MichaelKristopeit161 is operated by a pathetic individual attempting to steal the identity of MichaelKristopeit131, MichaelKristopeit105, MichaelKristopeit 89, MichaelKristopeit 68, MichaelKristopeit 45, MichaelKristopeit 28, MichaelKristopeit 10 and every number in between.

      Slashdot limits users to one account for each voice in their head.

    18. Re:so... by Anonymous Coward · · Score: 0

      dual arms even!

    19. Re:so... by Xachariah · · Score: 2, Funny

      You failed your DC 25 Human Lore check.

    20. Re:so... by Anonymous Coward · · Score: 0

      Now read this thread just after watching "Inception"

    21. Re:so... by L4t3r4lu5 · · Score: 1

      More so in the US. It's in the Constutution, after all.

      --
      Finally had enough. Come see us over at https://soylentnews.org/
    22. Re:so... by isorox · · Score: 1

      God bless the second amendment

    23. Re:so... by kevingolding2001 · · Score: 1

      Or just right click and 'open in new tab' which is what I did and why I didn't understand what blair1q was talking about.

    24. Re:so... by Ginger+Unicorn · · Score: 1

      man you're bored

      --
      (1.21 gigawatts) / (88 miles per hour) = 30 757 874 newtons
    25. Re:so... by ciderbrew · · Score: 1

      5ms? You need to upgrade those tiny screens.

    26. Re:so... by geoffball · · Score: 1

      More so in the US. It's in the Constutution, after all.

      It doesn't explicitly say that. But I believe the "Wifebeater" clause as interpreted by the Supreme Court does permit one to go sleeveless.

    27. Re:so... by MichaelKristopeit160 · · Score: 1

      ur mum's face is bored

    28. Re:so... by MichaelKristopeit119 · · Score: 0

      considering you'd need 54 tiny screens and a controller, 5 milliseconds seems, if anything, on the fast side.

    29. Re:so... by MichaelKristopeit121 · · Score: 0
      the truth (score:1) = overrated.

      slashdot = stagnated.

    30. Re:so... by Anonymous Coward · · Score: 0

      Wait! So now there's a constitution check also?

  2. The best part by Anonymous Coward · · Score: 0

    At about 6 seconds into this film i noticed the dude in the background looking at the computer. I'm pretty sure that's what every /.'er looks like while they're reading this post.

  3. Total Moves Not Moves/second by vortex2.71 · · Score: 1

    The total moves required to reach a solution is usually more important than the moves per second. While the robot is slower to physically move the device, it is probably moving the cube more efficiently than many humans. For instance, I can solve the cube, but do so by only moving one square at a time. More advanced solvers can move several squares simultaneously... a good computer will come up with the most efficient moves to solve the puzzle.

    1. Re:Total Moves Not Moves/second by Kunnis · · Score: 5, Informative

      The best solution for a rubix cube is always less then 20 moves (see http://www.cube20.org/ ) It averages around 18 moves for the best solution. That's 27 seconds to solve on average. Where this 11 second youtube video shows a guy solving a cube.. in less then 11 seconds. I can do it in about 60 seconds and I'm not very good. Also, before using computerized solutions, you have to know your whole solution. The manual solutions you can figure out as you go along... you can figure out your next steps while you're manipulating your current step.

    2. Re:Total Moves Not Moves/second by Ihmhi · · Score: 2, Funny

      I can do it in about 60 seconds and I'm not very good.

      I would think anyone who can solve a Rubik's Cube in around a minute could consider themselves good.

      I seem to recall an article about Will Smith and how he was taking up speedcubing. His teacher said a good goal for the average person - the point where they "get it" - is about 2 minutes. So you're at least twice as good at something as Will Smith.

      Incidentally, have you tried rapping lately?

      --
      Random Thoughts From A Diseased Mind (Not For Dummies)
    3. Re:Total Moves Not Moves/second by Anonymous Coward · · Score: 0

      I never timed myself and I never really figured out my own solution to the cube. A math professor I knew in college taught me a solution that he had figured out, that he had designed as being especially easy to learn (basically you unscramble one layer at a time). I think it tended to take 30-35 moves. He had much shorter ways to solve the cube himself. With his "simple" solution that he taught me, you could see the solved layers emerge as the solution progressed, but with the solutions he used himself, the cube looked scrambled almost all the way to the end where everything fell in place.

      Anyway, the "simple" solution he taught me really didn't require much thinking once you knew how it went. It was always obvious what to do next. So completing it in one minute would have been just under 2 seconds per move, not difficult at all. Therefore I think I could have gotten to that speed with just a tiny bit of practice. I never tried to go for speed though, and I've long since forgotten the solution so I'd have to learn it again before I could time myself. As I remember, it took him about 15 minutes to teach me all the steps, but of course I'd have to find him again and ask if he was willing to re-explain.

    4. Re:Total Moves Not Moves/second by migla · · Score: 1

      I think I can beat Mike Tyson.

      --
      Some of my favourite people are from th US; Vonnegut, Chomsky, Bill Hicks.
    5. Re:Total Moves Not Moves/second by Anonymous Coward · · Score: 0

      Are you sure he didn't publish it? I recall there was a very small book titled something like "The Simple Solution to the Rubik's Cube". The first couple of pages covered the notation, and the rest described a layered solution - top, middle, bottom corners, then bottom edges.

      Well, let's try google...holy crap! I'm not surprised that I found it, but rather that they're collectible, and one in new condition is being offered for nearly $1000. Unreal.

      - T

    6. Re:Total Moves Not Moves/second by Kunnis · · Score: 1

      What I acutally ended up doing was borrowing a co-worker's cube and messing with a bunch. Once you solve it a few dozen times, the action gets a lot smoother and it's faster to handle. I know the solutions I was doing were a lot more then 30 moves, but they were very simple to memorize, so I was able to do them quickly. I think I memorized a total of about 8 patterns, this is the site I used. http://peter.stillhq.com/jasmine/rubikscubesolution.html There are a few cases where I have to do the same pattern multiple times, but they are simple, so I learned to do them quickly without much delay.

  4. So now it's Android time? by Anonymous Coward · · Score: 2, Insightful

    I remember reading countless iphone stories which were completely pointless - anything done over iphone has been reported as some sort of nerd news on /. - and it was a good indication of rabid hype of otherwise ordinary electronic device.

    Now we see similarly pointless Android stories.

    This can only mean one thing - Android has arrived.

    1. Re:So now it's Android time? by WrongSizeGlass · · Score: 1

      Are you saying writing an app for the iPhone that only farts and makes other 'bodily noises' doesn't have any inherent value? Then you're sure not going to like "Chili Cheese's Raunchy Ringtone Collection". Best ringtones around (even better when they're combined with 'vibrate' mode).

  5. Greasing the skids... by CokoBWare · · Score: 3, Informative

    "Gilder says the Lego robotics kits can only manage around 1.5 moves per second, whereas human players can make between 5 and 6 moves per second, amazingly enough."

    Only if the cubes are greased well, otherwise they're stuck with wrestling the bloody things.

  6. Does it matter it was done with ARM? by adosch · · Score: 2, Insightful

    Maybe I'm just missing the excitement of this, but architecture aside, we know Rubik's cubes have predestined, mathematical (logical) approach to solving them, so really having any computational device (even like a microcontroller) can do that. I'd like to people fine-tune the robotic mechanics around turning and changing the cube, so it can start rivaling human solving speeds. I think that's were the feat would get a lot more interesting than seeing the next xyz-embedded computing device controlling another Lego Mindstorm.

    1. Re:Does it matter it was done with ARM? by fuzzyfuzzyfungus · · Score: 2, Interesting

      I'd be curious to know exactly how limited the hardware could be while still implementing the cube solving algorithm. Clock speeds are just a matter of patience, half the speed just means waiting twice as long; but there has to be an amount of RAM below which the algorithm Just Can't Be Done.

      As for the hardware side, I imagine that the guys who do really serious high speed CNC machinery could probably achieve better-than-human speeds, though the device would probably cost 100k and require the cube to be continuously sprayed with a coolant/lubricant fluid to avoid thermal or mechanical damage...

      Another interesting(and potentially cheaper; but with an exciting element of danger!) possibility would be if you allowed yourself to glue a suitably powerful magnet to each surface subcube of the cube, so you could then manipulate it without physical contact, other than one support point, by surrounding it with an array of electromagnets...

  7. Re:I don't play Rubik's Cube by darkrowan · · Score: 2, Insightful

    I'll take you as trying to troll, but the old adage "It is not the destination but the journey" seems to fit in this case.

    --
    AccountKiller
  8. Ah yes... by Oxford_Comma_Lover · · Score: 2, Interesting

    > thanks to a faster (1GHz) CPU, more RAM, and revamped cube-solving algorithms.

    I love how a few hundred million math operations per second is no longer enough for our phones, while most information in the human voice is under 8Mhz, IIRC.

    It feels a little like the math we teach our children is teaching them how to play with rocks in a cave. But they will never, ever, do what the computer is able to do already.

    Though they will do other cool stuff.

    --
    -- IANAL, this isn't legal advice, and definitely isn't legal advice for you. Also, Squee!
    1. Re:Ah yes... by Anonymous Coward · · Score: 1, Informative

      8kHz not 8MHz and that is the sampling frequency since POTS are 300Hz to 3400Hz

    2. Re:Ah yes... by icebraining · · Score: 1

      If by "math" you mean applying existing algorithms, you're right. But the fun part (and what mathematicians have always done) in Math is devising those algorithms. I'm happy to leave boring, repetitive work to machines.

      --
      Dilbert RSS feed
  9. Re:I don't play Rubik's Cube by WrongSizeGlass · · Score: 1

    It's a solved game. There's a known winning strategy. What's the fun in that?

    Not to mention it probably took a crap load of hours just trying to shave a few seconds off the process by revamping the cube-solving algorithms. There's no net gain by running this thousands of times just to try to break even. Why not try to get the people who staff my local drive thrus to put down their Androids so they can serve us faster. Now that would be an achievement.

    It is pretty cool that they can do this. I wonder if they can program it to knock the cube out of the human's hands?

  10. Anonymous Coward by Anonymous Coward · · Score: 0

    A human can do it in 7 seconds : http://www.youtube.com/watch?v=VzGjbjUPVUo

  11. Now do it with... by blair1q · · Score: 4, Funny

    A 3x3x3x3 hypercube.
    A cube where all the faces are slightly different shades of yellow.
    A cube where all the faces are pictures of people's faces.
    A cube made of jello cubes.
    A cube made of Plutonium-238 .

    1. Re:Now do it with... by marsu_k · · Score: 4, Funny

      or... a time cube!

    2. Re:Now do it with... by blair1q · · Score: 1

      While voice-dialing Domino's.

    3. Re:Now do it with... by ikkonoishi · · Score: 1

      A 3x3x3x3 hypercube.

      Here you go.

    4. Re:Now do it with... by Anonymous Coward · · Score: 0

      shouldn't that be a 4x4x4x4 hypercube?

    5. Re:Now do it with... by L4t3r4lu5 · · Score: 1

      No. 3 is the number of units along one edge of the whole face facing you. A 3 x 3 x 3 cube has 3 units along one edge, a 4 x 4 x 4 has 4 units. Very simple n+1 quantity where n is the number of parallel junctions it is possible to rotate a section of the face about.

      A 3 x 3 x 3 x 3 Hypercube would have 3 units along each edge, 2 parallel points of rotation. Including the edge you're having trouble picturing right now.

      --
      Finally had enough. Come see us over at https://soylentnews.org/
  12. There is one MUCH faster by gsgriffin · · Score: 1, Informative

    If you haven't seen this mind storm version, you may be impressed. It has done it in 2 seconds!

    Cube Stormer

    --
    jsut athnoer menagiensls ltitle psrhae for you to dcoede. Why do we wtsae our tmie dnoig tihs?
    1. Re:There is one MUCH faster by Anonymous Coward · · Score: 0

      Well done on posting a video without watching it first! A full solve took it 10.75 seconds. The shorter times were for simple tests.

  13. Cubestormer? by djlemma · · Score: 1

    There was an article about the cubestormer quite a few months ago here on Slashdot. It seems to be able to solve the cube much quicker, and it just looks a lot more sturdy. See video here.. So is it really that impressive that they did this using an ARM?

  14. Someone got carried away by vlueboy · · Score: 1

    Someone got carried away ...
    emulating those DROID platform ads.
    I'm hoping the real cube solver adds no fake robot sounds to its rotating arms.

  15. Re:I don't play Rubik's Cube by UnknownSoldier · · Score: 1

    Completely agree.

    Try to solve it on your _own_ without using anybody else's posted solution. Memorable 6 hours of my life.

    The same would apply to sudoku. Too easy, doesn't take that long, etc., but it helps keep the brain active.

    If the grand-parent is bitching about it, why doesn't he write his _own_ solver?

    --
    Inner Space, NOT Outer Space is the Final Frontier.

  16. Don't you mean... by Burning+Plastic · · Score: 1

    The Route of Ages?

    --
    [All Your Fish Are Belong To Us]
  17. Source / Instructions available? by landoltjp · · Score: 1

    Great demo! Will Mr. Gilder consider publishing the plans for the solving machine and the application source? T'would be cool!

  18. That girl by Anonymous Coward · · Score: 0

    That girl in the beginning of the video is SO CUTE. Who is she? I want to marry her :/

  19. Human Records by eulernet · · Score: 1

    The best average time for the human solvers is 8.52 seconds (with 5 consecutive cubes), and the best time for a single cube is 7.08 seconds:

    http://www.worldcubeassociation.org/results/regions.php

    Single 7.08 Erik Akkersdijk Netherlands Czech Open 2008
    Average 8.52 Feliks Zemdegs Australia New Zealand Champs 2010

    Getting 12.5 seconds from a single solve is not what I can call a benchmark.
    However, there is a larger progress' margin for computers than for humans.

    1. Re:Human Records by ToxIk_Waste · · Score: 1

      FYI, both these records were just broken by Feliks Zemdegs. That kid is a cubing prodigy. Also, for each average of 5 solves the best and worst times are dropped.

  20. Unfortunately... by JHaselden · · Score: 1

    Oracle have deemed this solution to be covered by copyrighted code inherited from Sun and are currently in the process of suing Lego. Because Larry needs a new dinghy.

    --
    www.abstractec.co.uk
  21. Nostalgia by Anonymous Coward · · Score: 0

    Wow, it's been a long time since I've seen anything about Lego Mindstorms. This makes me want to go find my old kit and start building robots again.

    Except I think it got thrown out a few years ago. Spring Cleaning, I hate thee.