Mecanum SwerveDrive

I think to myself, what do we need that we haven’t seen yet? More integrations with swerve drive. Specifically in the mecanum area. I have not seen teams using mecanum recently this project hopes to change that.




What we really need are some swerve drive modules with Omni wheels. Swerves are pretty maneuverable, and so are Omnis, why not combine them?



rollers are angled the wrong way smh

-1 pt


just gotta flip them around 180°

**Waiting for the STL files to 3D print this*:joy:

Not a true swerve… but close enough, no?

1 Like

I don’t get it.

A swerve with half the traction?

So I’ve actually wanted to try an omni-swerve at some point. I believe there are real advantages (and obviously disadvantages).

-The thing that initially got me thinking of this as a way to stop t-bone pinning. This would be like running an all-omni tank drive. It’s impossible to pin that way. I think this has some advantages for niche roles in games. With swerve, you could control how much force you put on a robot trying to t-bone you, and could potentially use that bump for a speed boost (if you were really good).

-Then I realized a more interesting advantage… theoretically infinite top speed (not practically infinite because of things like friction). Let me intuitively explain. Let’s say you have an h-drive with equal top speed in both X and Y direction. If both wheels move at top speed “V” then if you run both, you move diagonally at V*Sqrt(2).

So let’s put that on a swerve. You start out with all of your wheels facing forward. You top speed is V. Now as you start going, you start to make your wheel configure closer to 90 Degrees apart (like a robot with 4 omnis on the corners). Now your top speed is sqrt(2)V. Now keep going to almost a 179.9999 degrees apart. Theoretical your top speed would be insanely high (nearly infinite). I believe the graph is a Vsec(x/2) curve, but still need to work that out.

So in essence, you can get the advantage of a CVT if you do this.

Now I’ve only convinced myself that there are only 4 direction that this works on a 4-wheel omni-swerve. I am yet to try the math for the other cases.

Is this worth it? I honestly have no idea. Will I try it? Maybe some day.

There are certainly down sides like it being difficult to track the robot location on the field for auto (because there are not encoders on the rollers). You are also limited to a potentially less grippy omni wheel.

Also, if my math is totally wrong, or I overlooked something obvious, let me know. I’ve shown this idea to a few other mentor, and they are basically frustrated that there would be any advantage to do something so stupid.


I hate to break it to you… but Andymark already did this concept^… I will see if I can find the picture

^ as an April fools joke

Only thing that could make this concept more awesome is using Nidec motors!


FTC Team 3737 actually ran swomni in competition in relic recovery—they even had a cvt drive mode; granted, they ran it as an x-drive most of the time in competition as far as I know. There’s probably some video of it somewhere in, but I never actually looked for it, largely because you cannot see much.


Here’s another FTC team that’s done swomni:

And how can we forget about landroids in bowled over (that’s 2012!), with an x-drive with each module suspended independently and with 2 modules that can pivot to be straight:


Have you guys tried odometry/trajectories yet? If so how hard is it? This looks like it’s very maneuverable but would have a lot of slip.

I did once make an FTC-scale omni with wheels at 22.5°* to see if I could use rotation by 90 degrees as a shifting mechanism. I didn’t have any sensors, so I wasn’t able to get the system stable. I suspect that with encoders and a navigation module (or at least a gyro) this could be made to work. I did a few posts at the time, calling it Pakuni drive, after the furry humanoid race in Land of the Lost, who walked forwards but ran sideways.

* Actually 22.62°, as I based it on a 5/12/13 right triangle.


This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.