Help with my ID thesis in the SF bay area?

No problem - I believe an optical encoder would actually be better suited to your application, actually. Inertial measurement is much more expensive, and is generally only really needed if the entire object is moving - not a priority with your system. However, since you plan to be able to flip your device upside down, the parallax encoder isn't really a good choice - it would have to be one of the axis, and then the rotation would only work on one side. You could instead, make your own encoder - it's not too hard to do.

How I imagine your system is like this:

There's a central cylinder holding RGB LEDS, the Arduino, a single-axis accelerometer, the vibration motor, and a pair of IR sensors (the type you might use for line-following would be fine).
A hollow cylinder wrapped around the central cylinder and freely rotating, held in at the edges, with a black-white striped encoder pattern on the inside. The IR sensors would be able to read the strip to find change in angle, just like a normal quadrature encoder, but since it'll be held in at the edges, you will be able to turn the assembly upside down easily.