Mecanum/Octocanum Drive Robots

I have worked a bit on an octocanum drive design along with many other drive designs. What are pros and cons of octocanum? Personaly I havent seen it done much and I have never seen it done in Canada. Is there a reason? Will NEO’s make it more compelling? Are there any teams to watch competition videos that have an octocanum?

Pros -
Good mix of pushing and mobility
Easier to code then swerve from scratch

Con -
HEAVY
Complex to build

With Neos and the rise of available swerve examples/COTS, I just don’t don’t see the value in octocanum(or mecanum drives in general) anymore.

Octocanum is rare mainly because it’s ridiculously difficult and heavy, but I’ve seen teams use octocanum drives before. I don’t remember the video but there was a team playing with an octocanum.
Octocanum is also quite expensive, my team tried it in 2015 and were about $2000 in before we abandoned it.
Those are the main reasons it’s never used, but octocanum has one huge advantage: With a good driver, it’s got the best of both worlds: Manueverability and traction.

I hate to say it, but these days you’d probably be better off going with COTS swerve modules. They’re going to be lighter, faster, more maneuverable, and quite possibly cheaper.

I’ve only seen one octocanum in the wild.

It’s pretty heavy, expensive, and the set of teams who want to do octocanum over WCD and the set of teams experienced enough to pull it off well don’t really overlap. Most teams end up with the ability to switch between a mediocre tank drivetrain and a mediocre mecanum drivetrain, as opposed to a putting their time into building a good normal drivetrain in the first place. I wouldn’t say octocanum is that difficult, it just takes more effort to do well than many teams are willing to put in.

COTS swerve is great and all, but many, many teams are still vastly underequipt to tackle the integration, programming, and debugging side of things.

Just build a rock-solid, ultra-reliable WCD and call it a day. Or better yet, use the kitbot. Either way, use the time you would’ve spent designing, fixing, programming, debugging, and troubleshooting your octocanum/swerve/rocker-h/whatever drivetrain to instead train your drivers. It doesn’t really matter if you have the latest in swerve or octocanum or whatever, chances are you’re going to get outplayed by the team that spent their time practicing instead.

4 Likes

Another advantage some have incorporated into octanum and similarly articulated drive trains is that the wheel switch can be linked to a gear change. Most commonly, the traction wheels will be geared more slowly (for pushing power and off-the-block acceleration), and the mecanum will be geared faster. If you go this route, be sure to OP your cylinders, because you don’t want to have both sets of wheels on the carpet longer than necessary.

I don’t see NEOs changing the game of octanum relative to other drive trains, other than that it takes away some of the weight burden or conversely increases available power for the same weight. Also note that octanum is complex compared to skid-steer, it’s much simpler than swerve.

1 Like

We ran an octocanum for several years, and it was a truly beastly thing – I, and the kids, loved it, and it served us very well. We used 6" mecanum wheels and 4" aluminum wheels with rivet-on tread, chained together with an additional gear reduction off the 9:1 Banebots P80s so that we were going from a top speed of about 15 fps to a top speed of about 4 fps–and enough traction that we couldn’t turn in traction mode. This was before you were allowed to have six CIMs on a robot, and before the creation of the mini-CIM, so a four-CIM drivetrain was generally king.

The programming was easy–it’s just straight-up mecanum while you’re in that mode, and in traction mode it’s just tank drive; you switch the code when you fire the pneumatics.

Pros: Pretty easy to build, not terribly expensive, fun to build, fun to drive, really fast and maneuverable transitioning to really pushy and stable with the press of a button, impressive to sponsors, parents, and student prospects, helps keep your CG low, immune to T-bone pinning but really good at doing it back. Geared so low, the traction wheels let us push pretty much anyone (including tank tracks sideways) and served to stabilize us when scoring (either under defense or to just keep the thing from moving). Even playing massive beat-stick defense we never had an issue with breakage, but even so it was pretty easy to maintain if we did have a problem.

Cons: Preposterously heavy–like, unreasonably, stupidly heavy–and our attempts to make a lighter version compromised the awesomeness far too much on several levels; ghastly inefficient (you’re always driving eight wheels even though you’re only using four at a time–and usually those four are the mecanum which are already inefficient, you’ve got four gear boxes, and an additional chunky reduction off of each of those); takes up huge real estate at the bottom of your robot (though I’m sure this could be done differently than we did it back then, it’s still going to have a good chunk of space dedicated to it relative to, say, a kitbot drive or WCD).

So I disagree very much that it’s difficult to build or difficult to program, and a little that it’s particularly expensive for an FRC drive system (four mecanum wheels, four traction wheels, four pneumatic cylinders and the stuff to control them, four gearboxes, four motor controllers…if these aren’t things you already have kicking around then the initial cost is well higher than a standard drop-center–if you have most of these things on-hand and typically use pneumatics on your robot already, then it’s really not bad.)

Octocanum is fantastically cheaper, easier to build, and easier to program than swerve–we’ve done both, one with fantastic success and the other we almost pulled off–almost being enough to have our worst FRC performance ever.

1 Like

Swerve is better

Combine octocanum with swerve and you get a crazy dt

1 Like

We ran octocanum drive this year, and I’m not sure where the people who commented before me got their number. It is definitely cheaper than COTS swerve… $1200 vs over $2k
(If you make parts yourself. That’s different.)

Take a look at the VexPro Versadrop system. It works well if you don’t drop it off high surfaces… like level 2 Deep Space platforms.

In the end… you get mecanum… and you get traction drive… when you want it. It can be useful. Is it more maneuverable than swerve?.. nope. Your drive will also also be heavier… but we still managed to build a robot this year that did nearly everything this year bar lifting a partner with us while using one.

Last season we ran 8WD mecanum which is a newer idea that 1986 pioneered in 2017. Only a few teams have tried it. I think this worked really well for how we used it. I honestly want to play with it again with NEOs instead of the mini-cims we ran… for science. In the end mecanum can be decent in games like 2015 where there is no contact… or in games where you have some safe place to score… or defense isn’t worth it.

1 Like

What’s the point of 8 mecanum?

It drives smoother. I don’t have a way to quantify this with data for you, but I’ve driven robots with both and it is very noticeable. I found that you can strafe with better acceleration/speed/precision with 8WD over 4WD mecanum.

When I have a base with the drive again, I will explore sharing data that hopefully proves this. Maybe 1986 can show some. They ran it again this year if I’m not mistaken.

Anyone remember that post that mentioned 2834 having a great swerve defense this year?

The humorous thing to us was that we didn’t actually have swerve. Event Horizon has an Octocanum drive. I’ll have one of my design kids post some drawings.

OTOH those same kids do have an off season project working on swerve for next year.

2 Likes

Hey, I’m a design kid!
And the rumors are true we did have an Octocanum drive train this year.
We actually went through quite the rigorous design review even though we choose to go with the “pre fabded” versa drop system. total cost ended around $1200 not including the control system witch does require a pneumatic subsystem which can lead to increased cost if it wasn’t already part of your game plan.
we also happened to go through a couple of iterations this season, including wheel changes, motor changes, and gearing changes. we ended up settling on neos for drive and 16 fps on mechanum and 8 fps on traction. we switched from 4" OD 2" wide colson wheels to blue nitrial plaction of the same size after running tests and finding they could push better by about 25%.
overall I can only say one true negative to this drive train is space taken on your main side rails. elevator mounting was a real challenge.
Here are some screenshots from our CAD, the main changes to the standard configuration was the changed position of mechanum and traction wheel (this is to increase wheelbase) and the down gearing of the traction wheels from the mechanum. (this I see as a MUST, we ended up being quite the defense bot with this setup)
It also should be noted we did not use shaft collars on the drive train, we tapped the end of every shaft and used bolts with washers to close them off.

2996 built an octocanum in 2017 - I know they had a lot of issues mechanically with it. It’s a complex build no matter how you design it.

I assume that you essentially “clock” each pair of mecanums so that they’re out of sync with each other, which creates that smoothness. In the 1986 and 1102 implementations, it’s been done with 8 driving axles. Is there any reason you can’t do the same thing coaxially? Maybe one wheel on each side of a 2x1 with the chain running through the bar?

What’s so complex about making a mecanum drive with dropped-down traction wheels tied to them?

Probably the drop-down part, especially in comparison with a regular tank drive. It’s just less moving parts.

I think I’ll stick with my kitbot, thank you all very much.

3 Likes

Certainly not that I can see, especially if you’re using 6-bolt hubs and wheels with an odd number of rollers (most of the ones I’m familiar with). The inequality of torques around the robot center based on wheel placement would be less, as well.

–break–

I have a mecanum question, as I’m building a robot this weekend: Does anyone have load ratings or measurements or (failing those) anecdotal evidence about the axial/thrust load capabilities of the AM-2986 or other 1/2" hex flanged bearings? Mecanum wheels exert an axial/thrust force as well as a radial force on the robot, and I’m trying to figure out if these will be adequate for my design, or if I need to get thrust bearings. I’m looking at an FRC-legal weight and a roller CoF of around 1, so I’m looking at a static load around 30# and a dynamic load of about twice that before safety factors. I’m building this into an AM14U4 chassis, but totally in my own oddball configuration. I have put in a query with AndyMark, but I’d also like to hear from the community.

To answer my own question with 3946’s 2014 robot, “Buzz”: We ran 6" mecanum wheels at (IIRC) 10.71:1 off a TB-mini with one CIM per wheel. The gearboxes were mounted on the flange of the bent axle plate, and the axles were unsupported at the end, so it was sort of “West Coast-ish”, with possibly a bit extra “suspension” due to the mounting on an angle rather than tubing. The thrust forces inward towards the gearbox were actually taken up by the 3/8" R6ZZ in the housing. The robot was only about 120#, though. No issues, though we tried to avoid defense.