ScienceIsCool said:
The MIT motor is what is called a "pancake" motor because it's shaped like a disc. They've been around forever and I'm guessing what the MIT team was doing is finding ways to get sufficient torque out of it. Much, much more likely is that a hub motor is being used. My guess would be potted windings on the axle and the hub shell, run at very high voltage from a battery powered controller tucked into the frame. The only trick would be making sure you get a decent electrical connection from the dropout to the hub. It wouldn't be geared, so the setup would be low speed/high torque rather than something like the Vivax which can run at high rpm. That kind of motor is much easier to make. And I have no clue if you could build this hub motor to actually look like a "normal" hub.
John Swanson
I think the way to approach this would be to embed electrical contacts in each dropout, with the locknuts and axle completing the circuit, put the coil (windings) on the axle, with ferrous inserts or magnets in the hub shell. You'd need to electrically insulate the rest of the drivetrain.
Problem is, the hub on a pro road bike is skinny, and the axle has to be hollow and mechanically strong enough for a rider bombing down a mountain at 60 mph. That leaves very little room for a motor to produce enough power to make the whole thing worthwhile. If we start seeing pro bikes with bizarrely large diameter rear hubs, then someone should go put a magnetometer next to the hub shell - or toss a paperclip at it.
Here's a typical pro bike rear hub (a Zipp):