[BrianCloutier]

We've spent a while considering what happens when you add bumpers, and "I like the idea, but what about bumpers?" has been our biggest problem when pitching the idea of a roll cage internally.

Here's what we think:

1. Our CG is above (when the robot is upside down) the axis of rotation, so we will never be stuck on our top, we will always roll to one side.
2. When getting caught by the bumpers there are two different problems.
2a. You can get stuck with one set of wheels and the bumper on the ground. Just drive with your wheels and you'll be fine (Our CG will be low enough to give the wheels enough traction in this position)
2b. You can get stuck with the bumper and the roll cage on the ground.
3. We have an arm on one side of our robot that can put us into position 2a.
4. If we get into position 2b on the arm-less side of our robot there are only 3 ways to get out.
4a. Train our drivers to only go up the bump forwards and not backwards, so there is no chance of being on the wrong side.
4b. Have an alliance member push us over (Which will take very little energy) The only problem being they might be hesitant to go over the bump themselves to come and save us, and they're not allowed to come into the far zone if that's where we flip.
4c. Put our CG far enough forward that we always roll back to our arm side. A potential problem being we could roll back onto the bump and be back in situation 4b. Putting it too far forward also increases the chance of rolling when going over a bump to 100%.
5. There are few enough problems with the design to make the benefits worth going for. So we've chosen that this will definitely be part of our design. (It also stops balls from getting stuck on us.)