Just to clarify how my drive works a bit. My roller actually stays on the tire when the motor isn't in use. It's just barely on the tire but that's why I'm using a one way bearing.
Another side note... From what I've read about one way roller bearings you have to allow some radial play in the roller for them to work correctly. The EV Warror roller I use has bushings next to the one way bearing that are of a slightly larger inner diameter to allow for some radial movement so the bearing can lock up correctly. I've read some better explanations by some of the r/c guys. Here's one I found:http://rc.runryder.com/helicopter/t55473p1/?highlight=freya
If you scroll down and see where he talks about the Raptor one way bearing with it's somewhat larger bushings on the side you'll understand exactly how the ev warrior roller is made.
I suspect the people on here that have had problems are simply making a roller that can't float correctly. My current roller has around 3000 hard miles on it with the original one way bearing, so they can and do work.
As far as the other pivoting drives being more simplistic well, I agree and disagree at the same time. If you want a very very low power setup then yes. If you want any kind of usable power then no. These simple designs are getting more complicated by the day. Just read all the threads about custom electronics and servos to engage the roller and prevent choppy engagement.
Don't get me wrong, there's some cool stuff being invented on here but I like simple simple simple.
I also think the multi-diameter roller is a neat idea but why not just use a small roller and a higher kv motor? Wouldn't that accomplish the same thing with less stress on the motor?