Rubik's Cube Solving Robot (RDRK)

[connorbnelson]

I WANT TO BUILD ONE!!!

[SuperNiceworld]

Neparduodi to roboto?

[eAndrius]

@SuperNiceworld Tikrai ne ;]

[zilwinelis]

gaila tam jkm su c++ nemoko...

[rspure1300]

Saunuolis 5 =]

[torisiauliai]

Puikiai :)

[coleris007]

jai kada nors kandidatuosi i Lietuvos premijierus balsuosiu uz tave, bent įrodei kad nekvailas. ;-)

[TheSh4d0w1]

Šaunuolis! Sveikinu su kūriniu. Drąsiai duodu 5. :)

Ehm, gal galėtum dar sukurti robota kuris padarytu namų darbus už tave? ^^

[gyvaterojuje]

Šaunuolis

[FuturesF1re]

Lietuviai duoda i kaulus! :D

[Mac4s]

pzdc 5* iskart

[olliholliday]

this is absolutely awesome - but i'm wondering why it can't identify the initial configuration of the cube from the webcam instead of manually? it would be super-leet if it did it all itself.

awesome project though, nice one :)

[humblewizard]

This is so cool....I'm watching over and over. It's giving me a woody this morning. The robot never misses a hand-off. Nice work dude.

[alexfromnz]

whats the point of the camera feeding into the computer, when the human enters the colours for each side?

serves no purpose at all.

nothing new here, move along

[bonkute21]

nu cia geras ;]

[lauris47]

Nu mldc . Sukurk ta ka galetum parduot .

[Truthiness231]

I know I'm being technical and nerdy, but there are better algorithms for solving the cube that get the solution in less than 10 moves every time. Granted, the code would be a mile longer than it already is... ^.^

[eAndrius]

Yes, I know that. But LBL algorithm is actual for my needs. No, the most advanced algorithm known to humanity solves the cube in average 20 moves (at most by 21 moves if I remember correctly, same algorithm that RuBot uses, thought it's written by entirely different coders) and actually it wouldn't be necessarily longer that current code already is.

[Truthiness231]

Yeah not sure why I was thinking 10. Best I could find is 23, and 21 would be "leaps and bounds ahead of it" looking at the history...

In any case, not dissing your project. From one code monkey to another: gratz ^.^

[IAssemble]

@Truthiness231 - actually If you do the math to work out how many combinations of a scrambled cube exist and how many different positions can be reached in a given number of moves it is ralatively easy to demonstrate that some positions must need at least 17 moves because there are more scrambled combinations than there are positions reachable in 16 moves. I believe it has been proven that some positions exist that need at least 20 and it has also been proven that none need more than 22.

[Truthiness231]

@IAssemble It all depends on, as technical as one can get, if you count half turns as a single turn or two turns. But yes even counting half turns as a single turn, the upper bounds are beyond 10 ^.^

As of 2009, we know that the upper bounds have to be between 22 and 29 quarter turns, thanks to Tomas Rokicki. Amazing that there still is no definitive answer as of yet, but calculating 43252003274489856000 possible combinations using multiple algorithms for solution is pretty damn difficult.

[IAssemble]

@Truthiness231 - thanks for your response. Actually I believe Tomas Rokicki proved an upper bound of 22 (not between 22 and 29) So I believe that the current state is that the worst case is known to be between 20 and 22 moves inclusive (in HTM - half turn metric). BTW if you look at my channel you will see that I can get as technical as you like about Rubik's cubes ;-) And I should add that I do like your robot - very impressive!

[IAssemble]

@IAssemble - ooops sorry - I like eAndrius's robot... ;-)

[Truthiness231]

@IAssemble It's kinda hard to make out the facts from the Wiki page:

"In 2008, Tomas Rokicki was reported to have devised a computational proof that all unsolved cubes could be solved in 25 moves or fewer.[10] This was later reduced to 23 moves.[11] In August 2008 Rokicki announced that he had a proof for 22 moves.[1]

In 2009, Tomas Rokicki proved that 29 moves in quarter turn metric is enough to solve any scrambled cube" (iow, if you count by only quarter turns, the upper is 29)

[IAssemble]

@Truthiness231 - ah - sorry, I mis-understood, Yes, upper bound is currently 22 half turn metric or 29 quarter turn metric ;-)

[Truthiness231]

Oh yeah and it's 8 months late but I'd like to concur with IAssemble that your robot pwns ^.^

I played a bit with servos using MS' Robot Dev Studio back in college but never went as far as making anything as cool as this; you rock eAndrius!

[andriuss77]

Šaunuolis! :)

[ultra4477]

Amazing video , you rock ,thank you

[Avextinct]

Dude, now you should make it solve like those rubix cubes with pictures on em

[YOY0MA]

it could, you'd just have to asign each picture to a colour, see the robot doesn't need to check on it's progress, it saw where the different colours are, then it instantly knew which actions to do to solve it.

[turbomaxi2000]

WOW!!!!!!!!

[blinkys]

Tavo blogas nesmiges dabar? Jeigu taip, turbut del per didelio lankytoju skaiciaus.. (:

[eAndrius]

na rodos, kad nesmigęs

[simmerdown96]

this is really great!!! Keep it up!

[havefun2000]

How much time did the rebot take to solve it (in realtime)?

[eAndrius]

~10 min.

[l0lhuman]

Super.Taip ir toliau ;)

[TehDreamWalkeR]

jega :] !! super !

[paskraidysime]

geras, geras. mldc, negaleciau patiket kad dvyliktokas toki dalyka galetu padaryt .

[N1DE0N]

Andriau tu nerealus kiek laiko ir proto tokiam dalykui prireike as neisivaizduoju...

[Powerex199999]

Šaunuolis. :)

[Labviewer1]

Šaunuolis

[Arnis1mail]

geras

[gsjaeger]

This simply rocks!

[erKURITA]

Nicely done, simple and tidy. Though the hands are a bit slow, I like it :)

Thanks for sharing!

[SmileysDad]

What is the musical accompaniment? Very fitting. Nicely done.

[eAndrius]

Music: deep skydivers - momentum

Official site: deepskydivers{dot}com

[DoctorMelon]

Awesome! It looks great!