Revision history [back]

One issue is that this is open loop. @weasfas comment represents one particular (very ROS-y) way of closing that loop, you need to have something that's measuring the actual agular distance traveled by your wheels so you can derive how far the wheels have actually moved vs what you commanded it to move (less wheel slippage).

This is generally what we refer to as "odometry", which in short is just the actual measurements from some sensors about relative motion. This could be from wheel encoders, signal strengths, visual feature matching, etc. One of the first things you might learn in a control theory class is response time, just because you command a motor to go somewhere, doesn't mean it can instantaneously.

20 degrees is about a 5% error, in my book that's pretty reasonable for open loop control of this nature