A little earlier in the day I was reading a post by a robot hobbyist. He talked about his attempts to capture wheel information using a number of different approaches. One of his attempts involved painting lines on the inside of the wheels, then reading them with a LED/phototransistor. It worked, but since he was feeding the output back into his control circuit the resulting wheel action was pretty erratic. He tracked that down to the fact that he couldn't produce smooth, consistent lines on the wheels painting them by hand. He ended up paying some additional $$$ to buy a prepackaged encoder.
This evening, after dinner, I was reading some other robot construction articles and happened, purely by chance, to run across an article that Karl Lunt wrote several years ago for the Seattle Robotics Society describing his latest robot design. And, right at the very end of the article, was the solution to the wheel encoder problem I read about earlier the same day-
"The actual encoders consist of IR LED/phototransistor pairs that I bought at Vetco years ago. Each pair is mounted so it faces a 90-segment encoder wheel that I printed on my laser printer, then fastened to the inside surface of the wheel."
There is always a solution, and some one else on the planet has it...