[Unofficial] CubeStormer II - robot solves Rubik's Cube faster than human world record

[SpeedSolve]

Very cool.

 

[mycube]

I really like it! don´t know if someone asked before: whats the average of this machine? sub6?

 

[Stefan]

Humans don't have enough capacity to search very much (I assume - after all I don't actually know how Feliks Zemdegs' brain works! ) so why let the robot search if you're trying to compare like-for-like?

I'd say humans do try to look ahead beyond the cross to find good continuations/combinations. We're just not very good at it.

 

[Schmidt]

If you scramble a T PLL and have CS II solve that,
it would be comparable. I guess it will find the 11 move solution with the F/B moves that humans don't like, but it is still a T Perm.

 

[IAssemble]

I'd say humans do try to look ahead beyond the cross to find good continuations/combinations. We're just not very good at it.

I appreciate there is some look ahead but nothing like the millions of potential solutions that software solutions can try!

And let me say that I am totally in awe of you guys! I can only manage an average of about 39s myself

 

[IAssemble]

If you scramble a T PLL and have CS II solve that,
it would be comparable. I guess it will find the 11 move solution with the F/B moves that humans don't like, but it is still a T Perm.

Yes, the algorithm CubeStormer II uses will typically find an minimal turn solution up to about 11 moves and it doesn't care too much about awkward sequences

 

[IAssemble]

I really like it! don´t know if someone asked before: whats the average of this machine? sub6?

You're not the first but we haven't answered before because we haven't really measured it.

We're confident that it is sub7 and hope to record many times and determine an average when we demonstrate at ARM TechCon on the 26th/27th October.

 

[DavidWoner]

 

[MadeToReply]

Can you make the Robot mix it up to the most scambled form intsead of actually solving it?

 

[IAssemble]

Can you make the Robot mix it up to the most scambled form intsead of actually solving it?

CubeStormer II could be programmed to scramble the cube although I'm not sure what you mean by "the most scrambled form"?

 

[emg]

This is beyond cool. Congrats on your robot. It's awesome they're finally beating humans again, and lego no less!!

So now the question is, what do we have to do to get an official robot category? And what would the rules be? I've tried to get this discussion rolling a few times in the past but it just gets ignored. If it goes official, I just may have to try to make another robot (although at this point I'm really not sure I could beat you ;-))

-Evan

 

[IAssemble]

This is beyond cool. Congrats on your robot. It's awesome they're finally beating humans again, and lego no less!!

So now the question is, what do we have to do to get an official robot category? And what would the rules be? I've tried to get this discussion rolling a few times in the past but it just gets ignored. If it goes official, I just may have to try to make another robot (although at this point I'm really not sure I could beat you ;-))

-Evan

Many thanks. I'm glad you like CubeStormer II

It would be really cool if there was enough interest to create a category and have competitions in a similar way to humans! I am aware of probably about 30 other LEGO Rubik's Cube solving robot designs and of course there are some non-LEGO designs out there too.

I understand that Guinness World Records have an robot world record category - you can apply to them for a copy of their official robot rules if you're interested.

Good luck if you try to create another robot. It would be good for Mike and I to have some more competition! ;-)

 

[Stefan]

It would be really cool if there was enough interest to create a category and have competitions in a similar way to humans! I am aware of probably about 30 other LEGO Rubik's Cube solving robot designs and of course there are some non-LEGO designs out there too.

We have mosaic side events at world championships (and some other competitions?), maybe similarly robot side events could be held at least at the large competitions (world and continental). There should be several robots for an actual competition, and a big competition with its many cubers and prestige and media might be good incentive for you builders to come, right? If you'd like this to happen, you should get together and propose this to the competition organizers...

 

[Kian]

Until it picks it up itself it's not working with the same restrictions.

 

[IAssemble]

Until it picks it up itself it's not working with the same restrictions.

True.

And of course it is really a "blind" solve since the inspection time is included and the camera is turned off during the physical solve. So perhaps it should not be compared with a regular human solve for that reason either! ;-)

But if you want to see a robot BLD solve including pickup then you should see Mike's orignal CubeStormer (although I realize the displayed time doesn't include the pickup)

And please be aware that Mike and I are not trying to compete with human speedcubers. For us it's really just been a fun, challenging robotics project. We are both totally in awe of how fast you all manage to solve the cube!

 

[BadMeetsEvil]

Blew my mind. What method does it use? And how many algorithms does it use? Whatever - this is awesome.

 

[MostEd]

now imma waiting for 4x4 solver

 

[IAssemble]

Blew my mind. What method does it use? And how many algorithms does it use? Whatever - this is awesome.

Are you refering to CubeStormer or CubeStormer II?

Mike used Herbert Kociemba's two-phase solver CubeExplorer for his original CubeStormer design.

CubeStormer II uses my own two-phase software algorithm that I developed for Android Speedcuber. This software uses a lookup table for each phase but some of the information is encoded in a way that does not compare directly to human "algorithms". However it would probably fair to say that it has the equivalent of well over 100,000,000 "algorithms".

 

[IAssemble]

now imma waiting for 4x4 solver

Sorry, I know these have probably already been posted elsewhere...

and:

 

[Schmidt]

I like this "bad day" part