Incidentally, the large rear wheels in drag racers is an example of a variable torque wheel design:

http://www.youtube.com/watch?v=3Ug6w4ZjwVo#t=55

The automatic, passive adjustment is interesting, but these could just as easily use a conventional mechanism instead of origami. Is there some advantage to the origami aspect?

Potential for manufacturing processes, which are great at depositing in flat layers at very small scales. Structures manufactured flat can subsequently be folded into 3D shapes. It's a candidate method for making very small machines.

Wood's group (and possibly others - not my field) have even shown that pre-stressed parts can be built into the layers that cause the structure to 'pop-up' into it's 3D shape, or nearly, when released from its manufacturing scaffold. Neat.

Examples along with more nice things at http://micro.seas.harvard.edu/research.html

Another sometimes-cited motivation for folding robot parts is fitting planetary rovers neatly inside their spacecraft, or inserting robots into nuclear reactors through small service pipes.

Just a thought: Would it be possible to scale this technology up to the level of a car using folded graphene to act as the wheels?