Hi,
I've seen a working prototype of a "spherical electric motor" (on some university website I think, but I forget the URL), so it can be done. However, it wasn't the simplest of constructions.
As far as I remember, the concept was to use permanent magnets for the rotor, mounted on the inside of the "wheel globe" (the rotating sphere) together with solenoids (flat wound) for the stator, mounted inside a slightly larger hemisphere (you only need it to be more-than-hemispherical if you need to be able to lift the vehicle without leaving behind the "wheel globe").
The real trick, as far as I remember, was to make a controller for the solenoids, since this involves "sensing" the approaching permanent magnets, and then, depending on the direction of travel, activating the solenoids at the right time (and with the right polarity of course).
The result was a sphere (which I guess could be rubber, although they used hard plastic for their prototype) which could be rotated in any direction. I don't remember if the sphere would also "levitate" (ie. not touch the outer hemisphere when operating), or if they simply used some small casters to maintain the spacing between the sphere and the hemisphere. I guess the former would be possible, providing the solenoids and permanent magnets are strong enough (and that you are willing to spend extra energy for this purpose).
As far as I remember, the placement of the magnets and the solenoids wasn't very important, as the controller would have to "adapt" to the circumstances anyway, since no arrangement will provide a regular "pattern" for all possible rotational directions. Of course, a reasonably regular spacing is important. Also it might be important that the magnets and solenoids are placed in such a manner as to prevent a situation where all the solenoids are in the same position relative to the magnets (I guess this would be a matter of using two different "geometric patterns" for the solenoids and the magnets?)
Note that the prototype used a sphere roughly 10-15 inches in diameter, so it was a rather big "wheel". I guess that a smaller one could be made, but it would probably require quite small magnets and solenoids, which would still need to provide a decent amount of magnetism. Also the heat from the solenoids would need to be removed efficiently, as most permanent magnets are quite heat sensitive (ie. they will loose their magnetism if heated too much), but this might be a minor problem as the magnets are placed in the rotating sphere, while the solenoids are placed on the static hemisphere, so they aren't in very close contact, and not for extended periods (assuming the "wheel" is in constant motion and/or that the solenoids are switched off when standing still = not used to "hover in place").
Regards,
Inventus.
thethi.hs@gmail.com