Robotic Rubik's Cube Solver for Thesis Proposal

better to ask master benjo, hehe 20yrs in cubing, tsaka ang pagkakaalam ko, about algos in cube ang thesis nya before kaya kung kelangan mo ng tulong, i think si benjo kelangan mo.. kaya nya ipaliwanag sayo kung bakit ganun ang algo ng isang move unlike most of the cubers ay kabisado lang nila ang mga move..

@robert
hello, nice project..
tanong ko lang po kung anong robot gagamitin mo??
lego mindstorm po ba??? heheh.. kasi interested rin kasi ako gumawa ng robot solving cube pinoy version.. heheh.. kasi medyo may background rin ako sa robot maikng and programming using lego mindstorm.. and im capable of all the basic knowledege in solving a cube from 2x2 up to 5x5.. hehe.. interested talga kaso wala akong proper materials.. mahal kasi ng robot kit..

To kabhoom:
Well, it would be a big honor on my part to work or ask ideas with the best the Pinoy Speed Cubers can give. Even responses from people here on this forum is already a big thing for me. Could I ask for Master Benjo's e-mail address or any contact detail? Thank you for the endorsement, kabhoom.

To testing101:
I will be developing my own program, and be building my own hardware robotic design. I am not even familiar with the Lego Mindstorm that you are mentioning. But I guess it is from Lego the toy company right? I am glad that the a lot of people share the same interest as with me.

More power!

mga vids galung youtube mga kakaubang ga robot cube solvers sana makatulong to

Hello everybody. Thanks again for the vast interest regarding this topic.
Just would like to inform everybody regarding the updates for this project. As I have said, I am currently busy developing the program.

So far, my program is still capable of solving the Top Layer and the First Layer of the cube puzzle. That means, the software side of the project is now 30 percent done. Still have to make some major modifications though.

If you are interested of acquiring a sample copy of this program, I am very willing of sending a limited number of copies to people here. You can just give me your e-mails. Your feedbacks are greatly appreciated. But as I have said, the program is still capable of solving the the Top Layer and the First Layer of the cube.

Unfortunately, I will stop working on the project for one month or so. It is our term break and I will be back home (Cebu) for some vacation. And I will not be bringing my computer. Maybe I could continue the project next month. But if you are all interested, I could still send you the program.

The program's ability of solving the cube is still very limited and I have only trained it with few moves. So do not be surprised if the program will solve the Top Layer and the First Layer with a lot moves. Hehe.

These are my data so far for:

SOLVING THE TOP AND THE FIRST LAYER:

TRIAL TOTAL MOVES

TRIAL 1: 72
TRIAL 2: 70
TRIAL 3: 51
TRIAL 4: 134
TRIAL 5: 63
TRIAL 6: 86
TRIAL 7: 52
TRIAL 8: 68
TRIAL 9: 57
TRIAL 10: 72

The data proves a poor problem solving capability. But as I have mentioned before, the user can train the program by introducing your own techniques and store them to the program's memory. A learning algorithm, that is.

machine learning ba ginagawa mo?? nice.. anong language gamit mo???

@ Sir Robert, nicely done! When I started playing with the cube, I averaged more than a hundred moves all in all!

I am very interested in your software, can you possibly send it to me please? Maybe I can help with the algorithms to help shorten move counts.

My e-mail is benjediman@yahoo.com.

Thank you very much and keep up the great work, sir!

I am, too. However, I use a Mac. Still, I'll check it out on a Windows platform if such is the case.

My email is: graveyardshiftee@gmail.com.

Also, just a thought. I think Bryan will agree with me on this one: Instead of the first and top layer, why not try a 2x2x2 block? If this proves difficult, search for the Petrus method online. It will shed a lot of light on your program.

It might even totally use Petrus so that the total cube solution would be reduced to less than 50 moves!

ganun po ba petrus sandali lang e kung petrus man ang gagamitin edi intuitive lahat edi dapat marunong mag isip ung robot... i'm puzzled and confused

No, not really. Merong tinatawag na two-generators. Bryan knows that well, so tingin ko pwede. Isa pa, hindi completely intuitive ang Petrus. May algos pa rin yun. Yung transition lang from 2x2x3 block to last few moves ang kailangan ng kaunting intuition.

medyo interested ako pero kayo na lang but i'm fund of lego's.Lego minstorms are toys that you build and can move by its motors they call NXT's or whatever,Pero my vids sa youtube na Lego ang nagsosolve ng Cube just check it out. I love playing with them kahit ngayon just for fun pero pwede rin siquro kung ZB method kahit madaming algs yun they can be helpful kung tiyatiyagain:D

UPDATE:
Another development for my project. I have worked with my program last night and slept 3 am. And my program can now solve the Top Layer, First Layer, and now the Middle Layer, leaving the Bottom Layer and the Third Layer unsolved. However, the program solves these layers very poorly. It could even reach up to 300 moves, but sometimes 100 to 150 moves.

I am now collecting your e-mail addresses. Expect the software to be sent tommorow. Please give some feedbacks afterwards. That would mean a lot. Other interested people, please send your e-mail address, so that I could send the program all at once.

Maybe tommorow around 10 pm. I will send you the software copy as well as a manual on how to use it. But the program is still very rough, though working.

Thanks everybody for the interest and support.

To testing101:
Yes bro. This project is related to Artificial Intellegence. As I have mentioned before, I am making a software-robot capable of seemingly decision making skills. Learning process indeed. I am currently making it in Visual Basic, but it proves to be very slow. So maybe in the actual thesis proposal implemenation proper, I could code it in C++.

To benjediman:
Bro. My program is still capable of solving the Top and First Layer, and now, the middle layer. And with still the Bottom Layer and the Third Layer left, the program can solve the cube in more than a hundred moves. Sometimes even 300 moves. Hehe. But its ability is very much dependent on the techniques you have "taught" to the program. And maybe, not long from now, we can create some optimized solutions.

To Jome:
I have not tried my program to run in Mac. But if you have a Windows platform available, it would be very advisable. Thank you Bro, for another pointer. Maybe I will try to check that out, and see if that would really help. Whatever advices from people here are greatly accepted and considered. This project is becoming more of a team project. I owe a lot to this community if this project will really be materialized.

Wait, medyo nalabuan ako. Haha. Do you mean that it can solve the cube by using the Middle Layer as the first layer? That's pretty cool.

Hmm. Dapat talaga merong maganap na meeting or collaboration. Nakakatuwa rin eh. Actually, kung layer-by-layer siya, might as well use Corners First method. Or Waterman. Or whatever. The possibilities are endless.

Hoping to test it soon. Thanks. Will reply ASAP.

so you really have to incorporate a human-method and apply it for the
robot? optimizing one's method??... also try the thislewaite (im not
sure with the name.. but this is not a method for humans)...

again.. my email is marc_rendl@yahoo.com

siguro po pede ang keyhole method to lessen the moves... tas fridrich for the top layer.... pero kung pede petru para 1st robot to use petrus method

Hello guys.

I actually tried sending the program that I created last Friday. However, when I opened my e-mail today, I found out that it wasn't successfully sent. I am not sure, since I have not received any feedbacks from you guys.



I received this Failure Notice:

"Sorry, we were unable to deliver your message to the following address."

graveyardshiftee@gmail.com:




So I am resending it today excluding graveyardshiftee@gmail.com
I am only sending it to benjediman@yahoo.com and marc_rendl@yahoo.com
It is a ZIPPED file, and I also put some pointers and instructions on how to use the program. I just hope, that this will be sent this time. Pwede pakisend na lang sa mga tao rito. Thanks.


To jusay_mitch29:
Thank you for the YouTube videos you have posted. Ngayon ko lang napanood yung mga vids. Di kasi kaya sa bahay yung videostreaming. I don't have any form of high speed internet sa bahay. Nanghitch muna ako ng ibang computer. Hehe. But those videos are really a big help, though I am not yet thinking of the mechanical aspect of the project. But the third video I guess, is the simplest to build, I guess. Having only one rotating mechanism, and two other ejecting parts. The second one is very slow, obviously. The first one is just ok, but the mechanical design is complicated.

To Jome:
Sorry if I confused you bro. Sa pagkakaalam ko kasi, probably I am just wrong, that the cube is composed of five layers? Top, Bottom, First, Middle, and Third? Is this correct? Or am I just making my own convention. Hehehe. I am still new with the Rubik's cube, wala pa yatang two months.
For me bro, this is already a collaboration. But if what you meant, is to organize a meeting regarding this project, that would be very great. Though I cannot afford it.

To marc_rendl:
Yes bro. This program is a mimic of human brain's way of finding solutions to complex problems, like solving a puzzle.
There is no need to post again your e-mail. I am always taking note of all the PMs and posts regarding this topic. This is my topic in the first place, right?


Thank You Very Much for the continued interest.....
More Power..... and
GOD BLESS.....

Sir Robert,

I'm totally sorry, I did receive your email and had not the mind to reply! how inconsiderate of me.

But I've seen your program, and even if it has bugs as you said, I am still amazed by its capability of solving the first and second layer (We do not have 5 layers, sir. we only have 3, the top, middle and bottom layer. you must take your term top and first layer into one, maybe. just divide the cube into 3 equal parts ).

its probably asking too much, but will you ever publish your source? haha! maybe not, its a thesis anyways.

kung nagkakaroon man ng meeting kanya kanya ang gastos... as said in the posts before we should share what we know for free nd na manghihingi lang kami ng bayad kung alam talaga namin ung mga pinag sasasabi namin... pero sana ung robot mo uses petrus para 1st robot to use petrus nd sabi nila ung petru uses shorter moves to solve the cube edi shorter times kung ganun...

I only sent a copy of the program to two people here.
benjediman@yahoo.com and marc_rendl@yahoo.com

If anybody else interested of the tentative sample copy of the program, just ask from these people. I now have a limited access to the internet.

Maybe I could finish the rest of the program next month. As I have said, I am having a school break right now. But while having a vacation, I am now trying to do some sketches of the hardware part. Just some hard sketches on a dirty bondpaper. But maybe, once done, I could also share with you AutoCAD drawings of the hardware. The YouTube videos posted were really helpful.

To benjediman:
Thanks for enlightening me the correct convention of the cube. That means, I will be doing some minor modifications of the program. Another question as well, "what is really the correct color assignments of the Rubik's cube?". I am using a cube that I bought from National Bookstore that costs 150 pesos. And its color assignments are as follows: the blue side is opposite the white side, the red side oppostite the orange side, and the yellow side to the green side. I once bought a cube sold at the sidewalks costing only 40 pesos, and it has different color assignments. I am just confused with this one; and this parameter is an important consideration in making my program.
Regarding the source code. Of course, it is not asking too much. I can share it with you. But not now. Hindi pa kasi siya tapos. "You will never eat an uncooked fish". Right? But once it is totally done, there is no problem of sharing it with the people here.

To jusay_mitch29:
Regarding Petrus, I am still trying to study it. I am not that familiar yet with all the algorithms in cube solving. I am a total beginner in cube solving. I just learned some of the basic techniques, which I introduced to my program. Do not worry, I will put that Petrus as one on the top of the list. If I find it not very complicated for me to understand, why not.


More Power Pinoy Speed Cubers.....
and GOD BLESS.....

hmmm bat ganun ung nabili mo from nbs has japanese color scheme... lahat ba ng nbs cubes ganito??? anywiez... have fun in your vacation

The correct color scheme is, putting your Yellow on top and Green front:

Red - left
Blue - back
Orange - right
White - bottom

Kudos to Sir Robert!

ei sorry i didnt read your message.. i have not been using the internet since last week.. i'll b dloading the zip file later... im using a pretty bad dial up so i cant log in to yahoo..

Good day gentlemen. I wish all of you a fun summer. Enjoy the season. It is raining a lot here in Cebu.
By next month I will proceed with the software development. I am just hoping that by that time, you will still continue to give your interest and support. I am planning of finishing the whole programming part by June; so that I could focus more on the hardware design part which I believe to be the most complicated and more costly.

Something just came up in my mind. Just a thought, and you may just ignore it. So far, my best time of solving the cube is poorly 2.5 to 3 minutes. Is this good enough to be a member of PCA? Hehehe.

I was websurfing earlier. And I bumped with some videos of Tyson Mao on YouTube. I once saw him in Beauty and the Geek but he and his pair was eliminated in the early part. But I was not aware he was that famous in the Rubik's Cube world. Is he currently the best in the world?

I also tried to search for some videos regarding some cube solving robots, but there are only a few of them.


To benjediman:
I guess the cube that I bought from National Bookstore follows the correct color scheme; but the one I got from the sidewalks is not. Another question if you will not mind, the cube that I bought has colored stickers and due to wear and tear, it fades. Where could I find a "tiled" cube? If such thing exists, how much?


MORE POWER PINOY SPEED CUBERS!

the best is yu nakajima from japan(the last name gives it away )... its not based on how fast you solve a cube to be a member of pca... its just how dedicated you are

Good day everybody:

It has been quite while since the last post. But that does not mean that I already forgot this topic that I started a few months ago. Just wanted to tell everybody that I already finished the whole program.

The software part is already done, though still needs some very, very minor modifications. The program can now complete the cube.

Please post your e-mails so that I could send you a copy of the program. As I remembered, I sent a tentative copy (30% completion) of the program to some peope here a few months back.

I just started designing the mechanical hardware which I believe to be the more complicated part of the project. That includes: designing a robotic hand for the twisting and turning of the cube, developing the robot's vision system - web cam, and then interfacing all of those to the program.

Hopefully, I could finish the whole project, and could be able to present a working prototype, early next year, and maybe, earn my masters degree with that.

Thanks, More Power, and God Bless.