[Alex.Norton]

When my team has experimented with mecanum drive we found that placing the wheels in the same plane didn't necessarily help with angular drift. The problem is that even minuscule changes in floor height caused large changes in traction and the base would inevitably change angle. In addition we found that wheel slip was a big problem and directly measuring wheel speed didn't correct for drift either, even in a non competition environment such as our programing lab.

The best solution that we found was to measure the change in the angle of the base using a yaw rate sensor and implementing a PID control loop to keep the base facing the intended direction.

Of course we also found that a suspension to keep all four wheels in contact with the ground was helpfull. I could see were this would be a problem for a forklift and there was some debate on the team regarding if the suspension was necessary. I can understand where writing a complex control loop for the wheel could provide significant difficulty in this case but our team found that the best solution came from closed loop control of the four motors.