[Greg McKaskle]

This isn't RC cars, but it may be close enough, or it may spark some ways to get this implemented.

I think your biggest issue will be selecting sensors. I was in a competition years ago at Hanover Fair that took RC cars and attached a weak RF transmitter circuit. They had butcher paper on the floor with a race course marked out in black. Under the black tape they had a wire antennae snaking around the inside of the track.

Next to the course, was where you attached your computer to the wire leads of the antennae so that you could sense the RF signal strength and determine how far the car was from the inside of the track. You were also given the RC remote with the pots removed. You were able to send an analog voltage to the remote to control forward speed and front axle orientation.

This wasn't running on the vehicle, but you were only given a few hours to take one input -- not a great one -- and control two outputs to send the car around the track as fast as possible. In the end, this is the same as driving a car by looking out your side window and following the yellow line.

It was fun, challenging, but didn't involve targeting a tiny processor. Most teams of engineers had stuff going in a few hours. Of course most of the cars looked like they'd get pulled over and the driver would be given a breath test.

A related challenge is to do line followers. They can be done with any of the kits you mentioned, not to mention LEGO NXTs. The light sensor is easy to use and understand, and in the end the principles can be applied to other sensors to scale things up. There are lots of youtube movies showing results. Some of them are pretty impressive. This one is probably too easy, but you may be able to judge interest from the movies.

Greg McKaskle