For a sensor, he built a variable capacitor out of two semicircles of tinfoil mounted vertically, with a petri dish half-filled with water stuck in between them. When the wheel rotates, the water stays in the bottom half, which causes the capacitance across the tinfoil to change (because water and air have different dielectric constants). This change is then detected, and then used to change the frequency of a 555 timer chip, which is measured and translated into tilt by a microcontroller.
You could probably accomplish the same measurement by using an accelerometer in much less time, but I think this method is much more enlightening. I really like that the whole thing was made from scratch; it does a great job of demystifying how the sensor works. I realize that it was a second year engineering project, but it might be interesting to see what the dynamics of the system are (for example, how long it would take for the water to settle down if it was bumped?).
Does anyone think that it might be possible to design the capacitor differently so that it can sense which direction the tilt is occurring? You might also be able to use this kind of capacitance circuit to measure how much water is in a pipe, such as a rainwater gauge. Fun stuff!
Oh, and if you want to learn more about inclinometers, there is … wait for it … a whole blog dedicated just to them!