I apologize for any confusion from the presentation I posted - the picture is meant to show the pattern of contact between the wheels and the ground, as if you were looking from above but the top half of each wheel was transparent. I discussed this at length when I actually gave the presentation at the FIRST Conference, but perhaps I should have put more detail in the posted version...
UPDATED: I've edited the presentation to try to clarify the mecanum wheel orientation, and I've uploaded the revised version to both
http://www.firstroboticscanada.org/s...irectional.pdf and
http://www.chiefdelphi.com/media/papers/1836. (If you try to download the new version and it doesn't have an 'Updated' note on the first page, then it's probably the copy from your browser's cache - try using a different browser to get the updated copy.) Thanks to Ether for letting me know about this!