[seanwitte]

Part of the acronym soup for this group is PID. It's a shorted version of proportional-integral-differential control. The idea is to use a sensor to measure an actual output, compare it to the commanded output, and adjust until they converge. I think you can get by with a proportional term only the most of the time. There are lots of samples on these forums.

You can also use the technique to move from "relative" to "absolute" position. The basic example is a servo, which moves to a fixed position based on the input command. Instead of saying "go forward" or "go backwards" you call tell it to "go to position X".

The sensors commonly used are:

1) Gyro - analog sensor measures angular velocity. By sampling at a fixed interval integrating the output you can convert the value to a relative angle (think compass).

2) Encoder - device that measures rotational distance by being physically coupled to the thing you want to measure. As the shaft rotates it outputs a series of "ticks". They have a fixed number of "ticks" per rotation. By counting the ticks you can measure distance, and by accruing the distance over time you can measure speed. A quardature encoder allows you to determine both distance AND direction.

3) Potentiometer (pot) - variable resistor that outputs an analog value relative to the position of the shaft. This is useful for measuring a joint with a fixed arc (like an arm). For the arm application you can compare the output of the pot to the commanded value and drive the motor until the two values match.

Using this stuff you can write code to make your robot drive the way you want. For a simple tank-style robot I like to break it down into two commands: forward and rotation. Using a gyro or encoders you can more closely relate your forward/rotation commands to the actual performance of the system.

Good luck!