Wheel calibration

I noticed that my bot, when traveling in an undulating path, arcs very slightly to one side. I think this is because the wheel/motor resistance is greater in one direction than the other, at least for one of the wheels if not both. Since I'm not ready to implement PID control for motor speed yet (no odometry on board), I thought I'd try to minimize the problem by identifying the speed mismatch and adjusting in software by applying a constant modifier to each wheel's speed depending on the direction it was spinning.

To this end I replaced the wheels with CD's that had been drilled with a pattern of 16 holes around the circumference, and positioned off-board opto-interruptors under them to detect speed. The setup worked fine, but the over-all experiment was a huge disaster. The speed differences change wildly depending on motor speed and battery state of charge, so the measurements were all over the place. Oh well... He travels in a straight line just fine. It's only when making direction-reversing undulations that his over-all path curves slightly. It's neither important nor severe... It just annoys me.