pic: Can anybody give me advice on my Butterfly drive that I am working on?

I am making a butterfly drivetrain, and was wondering if anybody could give me advice on how to make it better. I’m not quite done, which is obvious, but the main issue I am having is how to actuate the rocking mechanisms that shift the robot from traction to omni wheels

Though a 4 speed robot would be cool to see, my first suggestion would be to eliminate your transmission and get your speed and torque changes by a difference in the diameter of your wheels. If you have a small traction wheel and a large omni wheel, you’ll get pushing power when in traction mode and speed when in omni mode. This will reduce complexity and give you more room to mount your pneumatic pistons for actuating the modules by making the modules smaller.

I’m curious whether he intends to do a 4-speed or not. The major tactical advantage of keeping a 1:1 ratio between the omni & traction wheels is to be able to change from one to the other at full speed. There is also another (huge, IMO) build advantage to this setup which should help answer the actuation question, and it is outlined on slide 3 of this paper:

I think I need more detail on how the drive train will be run before commenting further. For example, do you plan to use 1 serpentine belt or two belts per side?

I was just musing on the idea of pulling off a mostly-COTS 4-speed with this design.

While I agree that you could change modes on the fly, I don’t see a major advantage of this. If you’re changing to traction mode, you’re either trying to push or not be pushed. In that case, it’d be better to have the switch to low gear already built into the change of modes than to have to switch to low gear after changing modes. To put it simply, what advantage would low-gear omni mode give you? What advantage would high-gear 4-wheel traction give you?

From my experience, a major goal when developing a drivetrain should be to reduce the mental load on the driver. Having 4 possible drivetrain states for the driver to keep track of sounds like too much.

Based on the size of the pulley on the gearbox, I’d say he’s going for a 2 belts per side system.

Low-gear omni provides safe-zone agility while lining up on a precision game piece or goal, like 2011, 2012 or 2013 (2012 in particular).

High-gear traction provides faster straight-line movement since the 4 traction wheels on the corners make it harder for the robot to turn itself. This is tremendously useful in autonomous ‘drive straight’ routines like 2011.

Changing to high-gear traction at speed can prevent a robot from being knocked off course while still beating a defender to a point straight ahead.

On the surface I agree with your last point, that 4 modes is a bit much. Yet for the team to be successful with any “butterfly” drive, they will need practice. The modes I’ve described are fairly intuitive, so it wouldn’t be an issue for a team with a nominal amount of practice. I’ve driver coached lots of students over the years only a few of them would not have been able to handle a more intuitive setup of the modes.

From a mechanical maintenance perspective, 4 modes can be a nightmare given that there are 6 subsystems* with 6 actuation points in the current CAD above and 4 subsystems with 4 actuation points in the 624 design. I like that your design has only 4 subsystems with 4 actuation points and is effective yet simple, but I also know how my team does autonomous and driver training, so it wouldn’t be right for us. That’s a tradeoff the OP has to decide on as well.

*subsystem here is a contained system which is actuated (i.e. not the drive train frame or wheels which are not actuated). Each shifting gearbox is 1 subsystem, but a non-shifting gearbox is not.

aside from some of the points people have had you are gonna want to put some kind of support between the ballshifters because just having those angle bracket supporting the shifter won’t be enough on their own to be supported well

What’s the justification behind using belts to connect the two sides? One of the major benefits of a Butterfly setup is the increased mobility of the 4 Omni wheels when driven independently.

Another question, with the traction wheels on the outside, its possible that with a scrub-steer setup and 4wd; That the robot won’t be able to turn smoothly since the wheelbase is so long. It would essentially jitter (bounce) over the ground as the side wheels move in opposite directions. Not saying it would happen to this robot, since I don’t know it’s final weight; but my recommendation would be to place the omni’s on the outside and the traction on the interior. It’s what nearly every other implementation of Butterfly uses. See this butterfly implementation

Good Luck from your Cross-Town Rivals, see you at BB

From what I understand, a lot of teams with butterfly link the wheels on each side together so that they don’t loose power during a pushing match due to a wheel loosing contact with the ground.

Just an fyi, they had the traction wheels on the outside.

Oh, that makes sense; especially when up against the stronger drivebases.

Just an fyi, they had the traction wheels on the outside.

Suppose I should have made that clearer. Most teams I’ve seen run on the Omni’s the majority of the time, and the axle is run through that wheel because it’s the primary wheel. The axle placement on the traction wheel, at least to me, implied that the traction wheel was the primary.

We found that the pneumatic pistons worked really well (and were pretty small) to switch between omni and traction mode (plus gave the effect of low rider :D)

If you have any questions, feel free to ask me or the team and they can probably field your questions about it.


I am having trouble seeing how the Omni wheels get their power. Also, I didnt see a shaft for the omni wheels…

You could flip the orientation of the ball-shifter and puting on a double 25 chain sprocket inside the 1x2 and route chains to either module.

In Butterfly drive Omni wheels go unpowered, the use of them if is for drifting around.

Negative sir, they are powered and normally connected by gear or belt.

My mistake. I seem to reamber team 118’s robot having passive omni wheels on their Buttyfly drive. The designer of thus drive train is on my team so I can ask him about the unpowered omni’s.

He did say it wasn’t done. I assume the the final design doesn’t have floating omni wheels. (please correct me if I’m wrong)

That is correct. His focus shifted from this which, was a personal project, to designing a grasshoper drive with our team.

Robonauts did have Omni-wheels on their drive last year, but it wasn’t anything close to butterfly. They used the omni’s in conjunction with their WCD 8wd to perform J-turns and mad drifts.

Actually Robonaut’s drive was a subset of Butterfly called Texas Coast, the unpowered wheels are what makes it Texas Coast.