mecanum wheels - error in white paper referenced on Andymark web page

[Ether]

Is it my imagination, or are both Figure 2 and Figure 4 in the following link incorrect:

http://www.scipub.org/fulltext/ajas/ajas351831-1835.pdf

( the link appears on AndyMark's page at
http://store.andymark.biz/am-0083.html )

[meakerb]

Compared to the AndyMark spec sheet
http://lib.store.yahoo.net/lib/yhst-...lSpecSheet.pdf
Figure 2 is different, but not wrong.

On the AndyMark spec sheet, they state "Wheels in Positions 1, 4 should make X- pattern with Wheels 2, 3. If not set up like shown, wheels will not operate correctly." This isn't exactly true, it's just that the table that shows how to run the motors would have to change. The key, like the American Journal paper states, is to have the wheels arranged as "two mirrored pairs."
You do have to have two 'left' and two 'right' wheels though.

If, in Fig. 2, you take the two wheels on the left side and swap them with the two wheels on the right, you would have the same configuration as the AndyMark spec sheet.

In figure 4, they just have a different arrangment than their diagram, but it is still valid: they have two 'left' and two 'right' wheels (they probably just switched wheels around when their software created a spin move instead of a shift!)

So we have all the possible combinations covered:
AndyMark specsheet:
L R
R L

Figure 2:
R L
L R

Figure 4:
L L
R R

They're all valid, just different drive tables to get the desired movement.

Two interesting notes:
Neither Fig 2. or Fig 4. match Fig 8. which does match the AndyMark spec sheet.
If you look at Fig. 2 as a "top down" view, the rollers on the wheels will actually be the opposite orientation after they have rotated around and make contact with the floor. In this case, the force vectors are actually the same now as the AndyMark spec sheet.

[Ether]

Figure 2 is different, but not wrong.

In figure 4, they just have a different arrangment than their diagram, but it is still valid

Assuming Fig2 is a top view, how would you drive the 4 wheels so that the robot would spin in place, say clockwise?

In Figure 4, how would you drive the 4 wheels so that the robot goes straight forward?


~

[meakerb]

After giving this more than just a passing thought, I've determined that you are absolutely correct. There is only one way to orient the two 'left' and 'right' wheels and be able to properly drive in all six directions (forward, backwards, shift left, shift right, rotate CW, rotate CCW). That orientation is the one shown on the AndyMark spec sheet.

So I must eat some crow here and retract my original post. The only thing I got right was that the "top down" view of the wheels shown in Figure 2 would produce the force vectors as shown in the AndyMark spec, which would make things work. The wheel orientation shown in Fig 4 is just wrong.

My apologies.