pic: Experimental Octocanum/Butterfly Design

NOTE: Omnis are used in the picture. Obviously, mecanum wheels could be used.

This is an experimental butterfly/octocanum drive I’ve been working on. The center traction wheel only makes contact with the ground when the other traction wheels are activated.

Each module is individually driven, so that it can be used as an octocanum base. The center wheel is driven by a ball differential, and has a 3/16" drop. I am working on improvements to this design, including smaller traction wheels.

What are the speeds?

Why have the ball differential at all?

I can’t see a mode where you’d ever have the left side not all at the same speed (as you appear to drop front/rear modules at the same time).

That differential adds complexity, and is actually really difficult to tune properly.

A few of my rc race cars run ball diffs and it’s very much a feelings game.

He said he wanted to allow replacement of the omnis by mecanum wheels, to enable strafing without an additional mechanism. But this requires running front and back at different speeds, which requires the differential to power the middle guy.

But my question is why you need the central wheel. Is it that important to turn around your center on all traction wheels?

I think the intent of the ball differential is to leave the drive train open to being converted to an octocanum (decacanum?) drive rather than butterfly. To strafe, the front and back wheels would need to run opposite each other, which would create a conflict between the chains running to the center wheel without it.

I like the concept! Looks like a cleaner version ofthis. I’d love to hear from any 1322 members about their experience with this style of drive and the ball differential.

One alternative if the ball differential doesn’t work out might be to go with 6 CIMs, 4 on the back and 2 on the front, and only power the middle wheel off the 4 CIM side.

Also, 3/8" drop sounds like rather a lot.

Ah, overlooked that. Just looked at the picture mostly. Thanks

I wouldn’t run any drop on this kind of drive. 3/8" drop is extremely large for any drive - there will be WAY too much rock. The “standard” in FRC hovers around 1/8".

If my math is right, given a rigid chassis, centered CG, this robot should be able to turn with six wheels but no drop. It will be harder to turn than with a drop center but it still should be able to. Since you have omnis you can drop down, you can instantly turn about either end of the robot just by dropping a pair of omnis. Thus by making the drivetrain harder to turn in traction mode, you’ll have less trouble with defenders spinning you.

Some butterfly teams (most) primarily drive on the omnis. These teams omit the center wheel entirely, using traction as a “no spin” mode. Whether or not you do that depends on how you wish to drive the robot and the design goals of the drive, but it’s something to consider.

The drop’s that large because the robot is relatively long. It’s a parametric dimension that I threw a random number into, so it’s easily adjustable. Since the wheels are different sizes, and are always driven, you can’t drop one set of omnis alone. Even if they were set up to turn at the same speed, any drivetrain will be able to shove around a 2-traction 2-omni base would turn easier, but really wouldn’t help get out of defense.

It definitely helps with getting out of defense to drop one set of omnis. Instantly breaks a T-bone pin allowing the defended robot to spin around and pass the other robot. You’re right in that this trick only really works if the omnis and traction wheels spin around the same speed, but it’s a neat feature.

The other thing to consider about drop height is that once one set of traction wheels completely leaves the ground, extra drop is “wasted” as your base’s support polygon is already entirely contained in the other 4 wheels. I think once you get past 3/16" drop you’re in that territory.

Why not run belts/chains to the middle wheels only from one end This would allow you to turn front/rear wheels separately, allowing mecanum strafing, but would still drive all 6 when on the traction wheels. Seems a bit more simple than using a ball diff.

Unless you happen to be on a hill in the field while trying to strafe, when you’l end up doing who-knows-what when the traction wheels hit. And the last game I can think of with a completely flat field is Aerial Assist; before that I’d have to go clear back to 2007 (if you didn’t climb the on-robot ramps, which most teams did) or 1999 (if you weren’t a puck-climber, which most teams were). Every other game has at least had raised surface under the carpet. If you have a way of neutralizing the motion of the center wheel, you can at least get predictable results, which will be a lot easier to plan for.

And yes, I can think of situations over the years where strafing on a hill or irregularity in the field would provide an advantage.

You could do this, if it were a flat field. Actually, if it were a flat field your idea would be better. But, like EricH said, you couldn’t do this on a non-flat field unless you intend on only using traction wheels when on the bumps.

Sorry, looking at the original design, I assumed this was for a flat field, having very low ground clearance. I agree that it would cause weird issues if the middle wheel caught on a raised/inclined surface. Should have been more specific.

To be honest, any strafing drive besides swerve may behave oddly on inclines or on uneven surfaces just based on orientation and angle. It will work, but might not go exactly where you intended.