Sanity Check! Mecanum wheels...

Ok. Quick question: when you are looking at the robot from the top, should the rollers make an o-shape:

\ /

or an x-shape:

\ /
/ \


Another question: What have you guys done in the past to make the robot strafe straight?

Any help is appreciated!!! Thanks.

Shouldn’t matter, as long as all 4 corners are mirrored opposites. . . and you clue your programmer in to what you are planning. I’m sure one orientation may prove more stable.

Your wheels should appear as an x-shape when you look at the robot from the top.

We have found that even weight distribution, motor bias and wheel roller tightness all affect the ability the strafe straight.

Yes it MUST be an X on the top. If it an “O” on top… it will do everything BUT rotate. ( Actually it Will attempt to rotate but it will jump all over the place. Once you look at the vector you can see why.


Speaking from a team that had it’s team in an O shape and could not turn… YOU MUST HAVE YOUR WHEELS IN AN ‘X’ shape when looking down at the bot. Underneath, the wheels will look like a diamond or O shape.

Many of the pictures of mecanum wheels on the Internet are put on wrong. Several of the CAD files are also wrong. The wheels will go forward, back, and strafe but NOT ROTATE in a circle.

I’m still not sure how the orientation would effect turning. As long as the rollers are at a 45 degree angle, there should be no difference except the direction of wheel travel for strafing. Unless there is some variable that I am not seeing.

Same here. It should turn like a four-wheel in any wheel orientation.

If the “X” is on the bottom… the gripping vectors are facing each other so you are pushing in on two wheels and pulling apart on the other two. You have forces fighting each other and the rolling vectors don’t really do anything.
If the “O” is on the bottom all the gripping forces are outward at 45 degrees angles (from the axles) and opposing forces are canceled out by the rolling vectors.
It’s a lot easier to actually change the wheels and see the effect. I’ve actually done this with the practice bot and it leaves no doubt. You may be saying the same thing in different words since the key here is what’s at 45 degrees to the axle… the gripping vectors or the rolling vectors.


I finally drew the forces out on paper and realized the mistake I was making. It would turn, just not as easily due to the wheels not working together. So yea, looking down from above the X is the way to go. Thank you for correcting me.

Actually drawing things out helps sometimes I guess . . :rolleyes:



Our robots wheels look like the below but we are using omnis.
It strafes really well.

| |

We mix the two x values on the joysticks and stick it in the motors.

Our explanation for using the X pattern is that it is harder for other robots to push you around.

How so? As far as I can tell the static friction is the same either way your wheels are positioned.

yeah no matter which way you go you will always have two sets of wheels:
each set has two wheels parallel to each other, and the two sets are perpendicular to each other.

same amount of friction no matter how you look at it.


The two possibilities have very different turning behavior. Imagine the equivalent omniwheel drivebase with the wheels in an X shape instead of a diamond. You simply cannot turn yourself in place using that arrangement, but you also cannot prevent yourself from being turned by an outside force.

If the mecanum rollers don’t all point at the same place in the middle of the robot, as would be the case with a rectangular drivebase, you actually do have a slight ability to turn. But switching them to the other orientation will give much better control.

Good point, I hadn’t thought about applying a torque to your robot. I was thinking translational pushing, in which case the two configurations are the same.

If you’re confused about how the rollers make X’s or O’s, refer to the axis’ of the rollers. If the axis’ of the rollers make an X on top, you’re in good shape.