mecanum vs. omni

[Lil' Lavery]

Originally Posted by mateus

Maybe that would be a very stupid question... but I need to make it, so first of all i will explain the situation

Here in Brazil, no team has used Omni or Mechanum wheels at all... and i must say that my team has an old obsession about using it... but since no team here in brazil has used, is very hard to get them here, and we didn't talk so much about it before, some things are "dark" for us...

I would like to know the diference between Holonomic drive, Omnidrive and as i saw in some posts here... the Kiwi drive...
i mean about the movements of the bot... is there any difference or it's just the way the wheels and the system is designed??

We never heard about it before... for us using these kind of wheels would create just one more way to drive the bot... and now we see that we have 3 at least... we are confused...

I'd suggest reading some (or all) of the threads I posted earlier about vector based (holonomic) systems, as well as Ian MacKenzie's whitepaper (also posted on page 1 of thread).
A kiwi (killough) drive is really just another form of holonomic drive (just like a mecanum). A kiwi drive is a 3-wheeled variant, with wheels every 120º (as mentioned earlier). The vector math is basically the same, just with wheels every 120º instead of every 90º. A kiwi has a considerably smaller support structure, making it more prone to tipping if the bot is top heavy. The frame is also more difficult to construct if you don't have the proper tools/experience, as it requires non-right angles (although still not terribly challenging). However, with only three wheels, you don't suffer the problems of losing contact with a wheel (unless you bottom out, but that could happen to any drive), resulting in fluky driving. You also only have to invest the weight and space for 3 wheels/motors, as opposed to 4+ (you can really put any amount ≥3 of wheels in a holonomic system).
No solution is truly "better". A 4-wheeled holonomic, mecanum, and kiwi each have their advantages, but each have their disadvantages. The basic principles in them are all basically the same, and the real distinctions between these drives comes depending on how well a team can implement them.

[EricH]

One final thing to consider: the driver. Can your driver handle the extra maneuverability? IF definitely not, then stick with what works. If he just needs practice, do you have the resources to build a second frame (robot would be better, though), weight it down, and practice for weeks on end? If he can handle it without extra practice, give it to him anyway if possible.

Note: during testing in 2005, we noticed that strafing ability was impaired by holding a tetra about 2' away from the drive base. We just couldn't go straight sideways. We'd arc towards the tetra.

And yes, mecanums do tend to be slower sideways than front-to-back. At least, the first iteration does. It has to do with the fact that instead of having all the rollers work together, half of them are "fighting" the other half, so the robot moves more slowly. It looks something like this:
straight: --> --> ....sideways: --> <-- ....diagonally: --> --
........... --> --> ................. <-- --> ................... -- -->