Octocanum Drive/ Mecanum Suspension Designs?

As an off season project for our team we are looking at creating an Octocanum drivebase. We have experience with our own custom chassis, and have been using Mecanum drive for the last many years.

We have seen a few designs around (1086 / Blue Cheese, 3847 / Spectrum, 488 / Xbot, and 1540 / Flaming Chickens).

From my research, I can only find one example of an Octocanum that provides suspension when in the Mecanum mode (team 1540). It seems like with this drivebase we wouldn’t want to miss that opportunity. Does anyone have any experience with that? And can anyone point me to any CAD designs around that are set up like that?

Mecanum wheels apply force to the side when in use, so it is generally safer to be mounted to the frame rather than on a module. Plus, you tend to spend most of your time on mecanum, only using the traction wheels when you are interacting with the opposing alliance. In the event you are low on pressure, you can still use your main drive mode without issue.

Agreed on the point of spending most of our time on mecanum. I guess the trade off you made on your chassis was that suspension for the mecanum was not worth as much as the possibility of losing pressure and having the system fall back to the traction wheels?

2386 used mecanum for the first time this year and I too was concerned with suspension/ full wheel contact. In order to help keep our masts from swaying at the front of the robot, we used tensioned steel cables going from the front upper most point to the opposite rear low point. Although this did help with the sway of our masts, it also worked as a phenomenal way of ensuring all wheels were in contact with the ground. if the rear wheels off the ground, we loosened the cable lowering the wheels and vice-versa.

Should we ever go mecanum again, we will most certainly use this method of ensuring the wheels all touch the floor. To make it work you just need a rigid, tall super structure and build some flex into your chassis. Would highly recommend this method to others as it is both simple and easy to execute.

Alternatively, you could use a gyro to correct for the drift/inconsistencies and not have a suspension system. I know many teams that have mecanum this year who have no suspension, such as 1983, 2990 ect.

Yes, especially as it would turn the robot into a 4 wheel long bot(which can’t turn well.) The biggest issue that an octocanum has is it can’t turn well in octomode unless you have your traction wheels position inward far enough for it to be a wide base. Defaulting to mecanums fixes this issue.

Here’s a closeup of how 1540’s Octocanum drive works. It pivots on the traction wheel and uses 1" pancake cylinders to switch back and forth and to provide suspension. There’s a custom single speed transmission built into the 2x1 frame but you need not do that. There is no chain tensioning, we just CADed it to be the right distance and it’s been fine.

We started out with a variable air pressure system for the cylinder so that we could change how stiff the suspension was depending on how many toes we were carrying. We ended up dropping that feature because we needed the weight elsewhere. In the end, while we found Mecanum performance was better (more consistent strafing) with the suspension, our software with a heading sensor could do what we needed for this game. We actually don’t strafe as often as we thought we would.

For St. Louis we ended up dropping the cylinders to get the weight for a can-grabber. We still have the traction wheels but they are just used to go over the scoring platforms and are just fixed a quarter inch above the ground.

Be careful with this statement. Mecanum wheels absolutely requirement a certain amount of flex in a chassis system so that all 4 wheels maintain contact with the ground a majority of the time.

So either build a flexible chassis, or use suspension: one of the two.

A gyro can only do so much. With one wheel off the ground, you will run into a situation where the robot cannot correct itself and strafe at the same time. It will end up driving forward, backward, or sitting in place while the airborne wheel spins instead of sliding sideways like you want.

We ended up changing out our entire drivetrain to a slide drive at state champs. It ended up being lighter than mecanum and it always goes the direction you want it to.

1 Like

3309 used an octacanum drive in 2014 with suspended mecanum. We had the mecanum geared for 17 ft/sec and the traction for 4 ft/sec. It worked well but we had to make sure to conserve our air so as not to run out and drive on eight wheels going at two different speeds (yikes). Although we did well and that never happened in a match. Also looking back on it we probably would have used a 6 wheel west coast drive last year to maintain traction when doing turning maneuvers quickly (it depends on the game obviously). I will look for some photos and post it here if I find them.

Perhaps a crazy idea, but what if you did a “duodecanum” with the default being traction wheels?

This would have eight traction wheels, two on the same drive train as each mecanum wheel. The traction wheels are all at the same gear ratio as each other, and the mecanum all at the same gear ratio as each other.
Four of the wheels, one on each drive train, are located at the center front-to-back, and share the same axis of rotation. The other four are “outboard” of their respective mecanum wheels. The eight traction wheels form a standard 6-wheel drop-center drive chassis, with an extra pair of wheels. The four mecanum are off the carpet normally, but lift the chassis high enough that the drop-center wheel is off the carpet when the pistons are actuated.

What I’m thinking is that the center-axle wheels’ interactions with the carpet will serve to clutch the front and rear half drive trains together when in traction mode, but these drive trains will operate independently in mecanum mode. I’m thinking that the corner and mecanum wheels would be in a traditional octanum butterfly configuration rotating about the traction wheels, with the center wheels being driven from the corner wheels.

It would look sort of like this from overhead, with a-d indicating which drive train each wheel is on. [L] is a traction wheel, /L/ or \L\ is a mecanum wheel.


     [a]                             **

     \a\                             /b/


    [c][a]                          [d]**


     /c/                             \d\

     [c]                             [d]

Another edit: Another possibility would be to have ten wheels (decanum?), and a clutch on the center axle that engaged/disengaged the two drive trains on that side. Either of these systems would be programmed exactly like octanum, presuming that you’re already constraining the outputs so that the two left axes and the two right axes match each other when in traction mode.****

Having one of four wheels airborne sounds like more of a manufacturing problem to me, rather than a mecanum specific issue. There’s a crazy amount of teams using mecanum this year without suspension and very rigid (for FRC) frames, you don’t NEED it. The bigger issue (for the vast majority of mecanum teams) is the center of gravity distribution and strafing, which can either be corrected by a gyro, encoders and some code or not using mecanum.

Edit: I should clarify, varying weight on mecanum wheels is one of the larger issues with mecanum. Having no contact with the ground for one of the four wheels isn’t a problem that is specific to mecanum, it’s a manufacturing issue.

Yes, but it is often a manufacturing problem with the floor, rather than the robot. This year, as with many, it’s a manufacturing design. There are these things called scoring platforms that disrupt the planarity of the field. Even in that lovely land of theory where everything works, this will result in one or more wheels becoming airborne when the robot meets one at an oblique angle.

Snipped quotes above and referencing everyone else’s subsequent quotes as well here:

I don’t have any hard measured evidence, but it certainly seems like the drive-ability gets worse over time. I believe this is due to slowly degrading the chassis alignment over time from hard use (maybe due to chassis warping from hard bumps, post-competition demos where someone runs it into a wall, etc).

We did our first gyro corrected drive system this year, and it makes a difference, but cant account for everything. Now in our post-season, if we look at the wheels when strafing one always seems to be slightly off from the others; maybe it’s not resting on the ground with the same force as the others.

That’s what led me to inquire about the utility of the pneumatic suspension benefits of the octocanum drive. I agree we want mecanum to be our primary, higher speed drive, but its seems if all the parts are right there because of the octocanum, it would be a shame not to find a way to add the suspension to the mecanum mode.

But, there’s trade-offs with every engineering decision, and we will have to decide if the potential for a loss of pressure and reverting to traction, or some other octocanum failure is too big a risk to take. Or find a solution that doesn’t suffer from that drawback.

Sure, the scoring platforms disrupt the planarity of the field, but when would you ever need to strafe ON the scoring platform? The primary use of holonomic drives have been in alignment to feeder station and landfill. Expecting mecanums to behave normally while going over or on the scoring platform isn’t realistic, with or without suspension.

Even when you’re driving forward, mecanum doesn’t behave as you expect when a wheel is off the floor. The front of the robot will rotate towards the side of the airborne wheel. If it’s a front wheel that comes off the carpet, this will tend to rotate the robot so that it is more nearly parallel with the edge you are climbing, usually making matters worse.

And I can certainly imagine wanting to strafe across the scoring platform to cap a stack that was already built, or to pack the stacks in close to each other to leave room for more.

I just thought of something and I was wandering if it had ever been done before. Has anyone done a octicanum-like drivetrain with slide drive that can switch to 6-wheel drop-center? It would definitely push it with weight, but it might make it more worthwhile to go into traction mode for prolonged periods of time.

You lost me at “octicanum[sic]-like drivetrain with slide drive”. Octanum uses four mecanum and four traction wheels. Slide drive (aka H-drive) uses five omni wheels. 6-wheel drop-center uses six traction wheels. Would you mind backing up and taking another swing at this?

He is saying basically your idea for the sixwheel with drop down mec’s, except instead of octocanum use nonadrive.

We’ve used the Vex Mecanum wheels with Andmark Nanoboxes with No problems at all this year. We have encoders on each gearbox (3D printed bracket), gyro correction, and a very stiff frame. Wheels are set up to drive on to the scoring platform, and strafe to pickup totes in the landfill.

The Nanoboxes have very limited clearance between the upper mounting bolts and the CIM motor. We had to machine down the mounting bolt head OD to make these gearbox noise free.

Yay, I didn’t confuse everyone!