pic: Duodec-Drive (Octocanum)

Here’s the next iteration.

Changes since the last version:

-Added side plates to the cylinder mount and flipped it around.
-Moved the bolt pattern around, and increased all holes to a 1/4" bolt size.
-Changed to 9mm pulleys and added a belt.
-Moved the wheel as close to the supporting bearing as possible.
-Added some material back in around the cylinder connection on the module.
-Made the main cylinder mount set up for rivets.
-Made all module holes tapped.

How’s it looking so far?

My main questions are about the cylinder mount and the belt.
Does anyone see any problems with how I’ve gone about adding the side plates?

Also, does anyone know the thickness of the VexPro belts?
I defaulted to 0.1" because I couldn’t find anything with that information.


Any reason you can’t put a 1/4" thick gusset plate over the corner 2x1’s and just use a face mount pancake piston? Of even just nose mount the current piston. Everything would get a lot lower profile and you’d reduce your part count. (It would even be a great main frame gusset for that corner)


SDP/SI has dimensions for 5mm HTD belts here. The entire thickness (including teeth) is 0.15". Teeth are 0.082".

One of the cool things about this kind of drive, if it has the right configuration, is you get suspension for your Mecannum wheels for free. This insures all four wheels are in contact with the ground at all times. This makes a difference when strafing. I’d suggest flipping your design around so it pivots on the solid wheel. You’ll find the Mecannum wheels will work better if you do.

We used this kind of drive this year and in Bunnybots so got to measure this first-hand.

How does that setup work with the needed rotation of the module since the cylinder wouldn’t be able to pivot anymore? The change in distance isn’t that much, but that still doesn’t sound like the best thing for the piston.

I love the idea, I literally smacked myself on the forehead after reading your post… It made me feel foolish for not thinking of it first :o

We used a rigidly mounted cylinder in 2012. We positioned the cylinder to minimize the error and slotted the hole to make up the difference. Did not have any issues with it.

If you change the design to articulate the Mecanum wheels, you shall be intentionally and regularly creating forces along the axes on a cantilever mounted on another cantilever whenever you’re in Mecanum mode. Either you would have to “enclose” the module so that it’s mounted on both sides, or beef up both cantilevers beyond what you’d need for the current design.

Mecanum without suspension is OK if the field is fairly flat and the chassis is slightly flexible (e.g. aluminum C-channel or square tubing). If you have a good bit of driver practice, the drivers can even get used to significantly uneven loading. Woody, our practice 'bot for 2014 Aerial Assist had only 13# on one corner wheel and about 50# on the opposite corner. We practiced with it on a patch of parking lot behind the stadium’s concession stand; it was less than impeccably level. After a few hours driving, the operators had internalized the compensations. It did take a couple of practice matches indoors with “Buzz”, the competition 'bot, to unlearn those compensations. (Yes, Woody actually had a 2-by-4 wooden chassis and kicker and 2-by-2 pickup arms - video of driver practice here. We decided that the practice 'bot would be different than the prototype(s) and much more like the competition 'bot in 2015.)

In 2013 I did essentially this, and solved the problem by not attaching the cylinder to the module, it literally just pushed on a standoff, worked great.
(Spring pulled the module back up, reducing air consumption as well)


Our last iteration did something similar, but we pivoted the module on an axle that wasn’t shared with the wheels and used the weight of the robot to lift the module up. It also made maintenance easy since we could pull the entire module without pulling any of the power transmission stuff apart.

We ripped Aren off and did this too.

3928’s design from 2013 is so $@#$@#$@#$@# simple, that it honestly baffles me why anyone would do any variant of butterfly or octonum that isn’t a copy of it.

With Andymark’s new 4" mecanums, and 4" wide tube, an octonum variant is easy w/ his design.

It’s so darn simple.

Does anyone have a picture or link to this approach?


Doesn’t show the cylinder mounting concept that Aren was talking about.

In absence of finding a picture, that entire module sits inside a 3" wide tube. A 1-1/16" bore cylinder is face mounted to the top of this tube, and pushes down on the standoff above the traction wheel. there is sliding contact, and that is a nonissue.

EDIT: chassis posted here. Holes near ends on top face are where pancake cylinder mounts.

Uploaded it since it apparently didn’t exist on CD.

The piston shown is a 1-1/16" bore pancake piston from Automation direct, 1" stroke. It pushes directly down on a 3/8" steel standoff between the two sideplates of the module, and is not connected in any other way.

The module was spring return default to traction mode by a spring (not shown) attached to the small tab you can see behind the omni.
Pop the spring off and pull a single bolt and you could pull the module out.

Any questions about it just let me know.


The CAD for the module has been available for download on http://www.teamneutrino.org/seasons/ultimate-ascent/robot/butterfly/. You can see how the module works pretty well from it.

Mind blown. :ahh:

I had only ever viewed butterfly & octocanum drives as a curiosity, but now I’ll have to reconsider!

That is an amazing design! We might be stealing this for next year :smiley:

I would recommend pivoting on the traction wheel unless you have some serious side support. (See Aren’s image where the module is snug between the channel.) We had a lot of problems on the 2010 148 drive with this. Your traction wheels will stay put, but the modules will flex sideways under sideloads. We used chains and had issues with them coming off. May not be such a big deal with belts, though.

To elaborate on this, the sideplates of this module were .125" thick steel plate, and there were also two pieces of adhesive backed HDPE between the module side plates and the inside tube faces to send the forces straight into the 3x3 tube.

The main reason I chose to pivot on the omni wheel was to keep the CIM hard mounted, rather than swinging with the module (single stage of gearing to the omni wheel).