[Jon Stratis]

I would say it's highly dependent on the task.

There are complexities with crab/swerve that aren't present in the others - with a short build season like we have, it may be too complex to accomplish (when you consider everything else you want to accomplish). Then again, they provide more pushing power in any direction than the others - is pushing power important enough to accept the additional risks that come with the complexity of crab/swerve? Do you have the weight available for crab/swerve (as it is typically heavier than other drive systems)?

Having Mecanum wheels on the corners works great, so does having omni wheels on each of the 4 sides. But if you need to go up a ramp, an omni wheel setup like that isn't going to work, while mecanum will. And just taking the omni's at 45 degrees to the frame in the corners is basically mecanum, only with the added complexity of needing to mount everything at 45 degrees to the frame.

So before you can really get good answers, you have to define the task and the associated parameters. Each system is good, but different systems will excel at different tasks and with different parameters. One of the key challenges with building a robot for FIRST isn't identifying a single "best" system and perfecting it to use year after year. It's in identifying what the best system would be for a given task and working in a very short time period to perfect it.