Quote: Original post by kirkd
I'm not familiar with plan space search methods. If you have a link to what you would consider to be a good review, I would greatly appreciate it. I haven't search CiteSeer or Google Scholar yet, but I'd certainly take your recommendation.

-Kirk

A good review... mmm, not off the top of my head. However, I do know a couple of people in this area... I'll look into it for you. If you're looking for articles though, I'd search the Operations Research literature, rather than the AI literature, since this is essentially a optimisation problem, but in operator (action) space, rather than state space (which most AI people work in). I'll see if I can locate some specific starting points for you.

Cheers,

Timkin

Excellent! Thank you, Timkin.

Quote: Original post by ThreeDParadox
I did get a couple laughs out of it though, like "anti-clockwise". I think im going to pass on writing my own rubiks cube solver, for now at least. Ill wait until i get some more AI under my belt before taking on such a task. Thanks for the help!

I'm guessing your american, because it didn't say counter-clockwise :)

We say anti in the UK :)

Quote: Original post by kirkdFor humans, the main method is to solve one face with the edges correct, solve the middle layer, solve the bottom corners, and finally solve the bottom edges. This takes ~175-200 moves to do, which is FAR from optimal, but is easily remembered and interpreted by humans.

I agree on the method, but the number of moves is way out. I use a very slight variation of that method (I solve the bottom edges before the corners), and it comes in at around 100 moves. I know it's still inefficient, but it can still be done in about 55 seconds or so. Just thought it was worth a mention.