The motors have a slight bias in their windings that makes them spin faster in forward than reverse. For the CIM’s it’s hard to notice. 2004 and prior it was a big deal, as the drill motors had a significant forward bias, and the robot would drift fairly badly if not compensated for.
For more than you probably wanted to know about how motor performance can depend on direction of rotation, read this thread.
We avoided this problem this year (after deciding on kickoff weekend to go with mechanum drive) by building a drive-train quickly in 1 week while software worked on the code. Then we let software do as much as they wanted in terms of code testing to make sure that it would work. The early production time frame for our drive-train made certain that that we could switch back to 6 wheel if software couldn’t get mechanum code to work. As it turned out, the saturday after kickoff, at about 6:00 pm, about ten minutes after integrating code and drivetrain, this was our result (link at end of post). Pretty amazing result for the turnaroud time. We left it with software for a few days so they could tweak things, and the design/fab team continued on with the rest of the 'bot. This video was taken before we changed gear ratios for a considerably faster robot. We figured we’d start slow, and work out the kinks with the code, then speed it up. And sorry for the bad quality, it was taken with my cell phone.