This one is cool because it can do 2x2x2 up to 5x5x5.
A couple of things have got me thinking about these creations. The first is that the builder uses an old Vision Command camera. I happen to have one of those, and it is stored in a plastic box at the top of my Lego shelf. I would love to get that involved in some of my robots. Does anyone have any experience in hacking that thing?
The second thing I noticed is written on the description of the robot.
The application uses a generic algorithm that can solve a cube of any dimension limited only by RAM and time. It has been tested with software models of cubes up to 100x100x100.
I suspected that this was the case. I wondered if the solution algorithm changed every time the dimension changed. What this means to me is that there is little difference between a solver that does a 3x3x3 and a solver that does 33x33x33. Both are equally impressive, but the software is essentially the same.