[J_Miles]

Perhaps I'm simply biased, but as a driver I loved driving a robot with mecanum wheels. Our 2011 robot was built on a mecanum drive and won the 2011 Ann Arbor District competition, the 2011 Troy district competition, as well as two local off-season competitions in Michigan. It was definitely a machine to be reckoned with in spite of what many would consider to be a poor choice of drive train.

With that said, there are a few advantages to mecanum drives, but I think they're quite context-based. 2011 seemed to be an excellent year (if there is such a thing) to use the system because of: A) The totally flat field, B) Sizable protected scoring zones C) A shallow protected zone (making the ability to strafe quite useful). For this reason an omnidirectional drivetrain seemed the way to go, but we chose a mecanum drive for its relative simplicity and our previous experience with it, and our year was spectacular because we were able to put lots of effort into our end-effector and other mechanisms besides the drive, and to build a solid mecanum drive there isn't a lot of fancy programming necessary besides speed control.

Mecanums have the advantage of being able to immediately change from forward & backward motion to sideways motion without having to worry about the wheels turning. Driving on mecanums also means that shifting transmissions are sort of irrelevant (as you aren't going to get much pushing power from mecanums and, as such, shouldn't need a significant boost in torque), and because the rollers allow for low-resistance turning.

Surely there are sacrifices made when designing a mecanum drive for a robot, but I still argue that they are a solid option if you want a simple omnidirectional drivetrain and are willing to live with the consequences.

Also, as an aside, the "drunk drive" (as we called it) that comes with mecanums significantly changes the driving experience, and I personally found it a lot more fun than driving the standard tank drive we used in 2012.