So a butterfly/ octocanum drive is where you switch between mecanum and traction wheels, right?
I was wondering what the typical weight and complexity of these systems was. This would be weight per module or, preferably, weight of bellypan + drive base + modules. I have only found one weight listing for an octocanum module, and I think many teams keep the mass under 11lbs.
How long does this take to design? Does this need a particular CNC special part or can it be fabricated with a waterjet/ sheet metal shop?
Any information on these drives would be useful, as I’m looking into different drives for a summer project.
Assigning names to a “design” is so limiting. That said a butterfly drive conventionally switches between 4 omni wheels and 4 traction wheels.
I would encourage you to sit down with your team and figure out a list of objectives you would like your summer project to fulfill. Once you’ve done this I would encourage you to use the chiefdelphi search function to read up on butterfly drives from older threads. Once you understand your design objectives, have an idea of where a butterfly type drive falls on that, and seen a number of designs other teams have already made you can properly assess if you should try to design and build one or if your time would be better spent working on something else. When done properly off season projects can be an extremely valuable resource both in terms of populating your team’s design shelf and teaching students. The most important thing is to have a full understanding of why you are building what you are building and how you are trying to improve upon what you have done in the past, have seen others do in competition, or online.
Cheers, Bryan
Actually, right now it’s just me looking at drive train ideas; other people have AP tests to study for right now and just a lot of schoolwork in general. I want to have a list of detailed CAD designs to show the team later once we start thinking about summer projects, so that we can make an educated decision about what we want to try.
I’ve been cadding only swerve drives and shifting gearboxes up to this point, but I was wondering what I should aim for in terms of weight and size if I CAD a butterfly later.
I want to stick with swerve personally, but the problem is that the programming is very difficult. With a butterfly drive, there is already a lot of programming released, not to mention the programming overall is easier.
As BJC said, butterfly is [typically] switching between omni and traction wheels; while octocanum is [typically] switching between mecanum and traction wheels.
The main difference between the two are choosing what wheel is opposite of the traction wheel on the pod. While I have not built one, the drives should be slightly heavier than a conventional drive. Majority of these setups use a single gear reduction to the first wheel and then another reduction via chain or belt to the second wheel. This makes up for the weight “missing” from gearboxes.
There are some good examples out there on how they have been constructed in the past. 148 made this type of setup popular in their 2010 release video and their CAD files can be seen on FRC designs and you can see another iteration in their 2011 robot. Another great example is 3847’s robot from this year. It was documented fairly well in their build blog. Both of these teams are more than willing to help answer further questions, but these should give you a good starting point. There are plenty of other robots that use these drive types that I didn’t mention, these were two I could get my hands on quickly.
If you design an octocanum or butterfly, be ready to spend at least two or three months to go from concept to working prototype. If you use bent sheet metal for your modules you need a water jet, if you use 1/4" aluminum plates, you need a good CNC. Do know that they can be much heavier than other drivetrains depending on their features, so they aren’t suitable for some games.
Though I doubt it would work with mecanum, we used a new format of the “Tex-Coast Drive,” (Grasshopper) essentially combining butterfly with a West Coast drive. We don’t CAD nor do we have a CNC machine, so simplicity and ease of construction was something we needed. We built the first prototype and the competition drivetrain both in the same six week build season. It weighs just a bit more than a standard 8WD because of the pneumatic actuation. Here’s a link to what I mean: Grasshopper Drive System
We went from concept to working prototype in about 3 weeks during build season this year. We didn’t do any prototyping or design on that type of drive train before build season started. Do your research and don’t be afraid to ask for help. Our drive train wasn’t perfect but it did very well and we had to perform very little maintenance on it all season.
Also as noted about it’s called Tex Coast Drive now.
There are trade offs in every train and depending on the features you put in it can be lighter or heavier.
I had a great opportunity to talk to a couple of people on your team about your drive train, including manufacturing and maintenance. The insight was invaluable. If we decide to do it next year and improve upon it, we’ll definitely send you a heads up with any of our new findings. From what I saw it was the best blend of simplicity, maneuverability and immovability. Who knew it was based upon a concept that was put out 3 years ago by the Killer Bees?
The only thing about the whitepaper I don’t quite understand is the ‘elastic tensioning’. Could you explain?
As RobotDoktor said, these drives can easily be very complex and heavy. We used it on our competition bot this year, naming the drive base the Dragonfly (And we won Excellence in Engineering at CMP for it). It switches between colsons and mecanum wheels, with each module weighing about 9 lbs. The entire thing with the electrical board, for us, was about 100 lbs packed into 6-8 inches of height. We came so close to the weight budget that we had to cut wires as short as possible just to cut down on weight!
A couple years ago our team built a ‘slide’ drive that used two omni’s that were in the middle to translate us and four omnis at the corners to drive. It was terrible, but that’s where I first learned of these concept drives.
At the time it occurred to me that you could use something equivalent to a starter bendix drive to push-down the translating wheels only when you need to strafe. In a starter bendix, it moves forward to engage your flywheel using the rotation of the motor–Simple, no space-hogging pneumatics on the frame. I have’nt tried it but I’d be interested Yes, a bendix only moves foreward on one rotation, but I think you could use the concept to design a bendix-in-a-bendix and make one that moves down on both rotations.
And then it occurred to me that this might be better with four-wheels for strafing and four wheels for regular driving . Then I thought why do we even need omni or mechanam wheels for this?
So why does an ‘octonum’ or any of these others even need translating wheels? If your goal is to strafe, can’t you just do that with ‘good’ wheels?
This sounds quite similar to how the Robowranglers deployed their middle strafe wheel this year.
Here you go. And thanks, it was an experiment for us and it turned out pretty well.
This year we developed and used a hybrid of an octane. we went with six wheels instead of 8. We used 4 2-stage vex transmissions geared 3:1 and 8:1. We also used a 12 tooth sprocket from the transmissions to mecanem wheels, and then a 22 tooth sprocket on the mecanem. We then moved to a 44 tooth sprocket on the 6 inch wheels. We built two pods for the mecanem and traction wheels duos, and then ran a bar connected to each beneath the central frame of the robot. A 3 inch piston pushed the bar down in the middle of the robot, engaging the traction wheels and lifting the other mecanems in the pod off the ground.
See link for a picture.
http://www.chiefdelphi.com/media/photos/40492
We welded the pods using 1/4 inch bar aluminum. Our frame is made of c-channel. It was not the lightest of drivetrains, but we still fit everything else on the robot with not too much worry about weight.
We could push 90% of the robots on the field, and we could hold our ground against the other 10%.
I don’t remember the terminology too well, but wasn’t the 2010 robot a Nonadrive, since it had a ninth wheel that was actuated down to allow for holonomic drive? Their 2011 robot was iirc their first true butterfly after they realized that the ninth wheel wasn’t required for basic strafing.
I talked about the history of 148’s articulating drives, and the evolution of the “Tex-Coast Drive” here:
http://www.chiefdelphi.com/forums/showpost.php?p=1366499&postcount=15
You are correct in the fact that in 2010 it was nonadrive; however if you ignore the 9th wheel (omni in the center) it is an iteration of the drives being discussed.
I think terminology, albeit important, can sometimes limit creativity.
Thanks for all the replies. I especially enjoyed the Grasshopper drive (I saw that a couple weeks ago) because it seemed very lightweight @ 30lbs. Also, it didn’t look like it would take too much effort for even an inexperienced team due to the fact that the amount of lift on the wheels could vary quite a bit.
So it looks like these modules are a little on the heavy side at around 8-12 lbs, maybe 6 or 7 optimized. It looks like these could be pretty useful to us next year in leiu of a heavy swerve. Thank you so much for all the input!
Our octocanum modules are 5.75 lbs. each. The pneumatic cylinders driving each pod were mounted to the frame and I don’t have the weight of each of those handy. We used four, but you could as easily use 1, 2 or 4 as your frame allowed.
Our first octocanum implementation in 2011, itself the first used in FRC, was made entirely in-house and with no advanced machining techniques. It was designed and built in just a few weeks and while heavy, it was very reliable and effective.
Our 2014 implementation is made with sheet metal and that helped us to realize a lot of weight savings. We relied on our experience from 2011 to design and code the drive quickly.
Wow, that’s super light! Even with cylinders, I can’t see that popping up to more than 7lbs.
We have a sheet metal guy, but CNC is up in the air right now (our local college can only machine on weekends and they like the manuals).
I’ve been looking at octocanum drive trains as well and come up with several designs so far. I found this thread of 2583’s design very useful, and there’s information about calculating whether or not you get enough force out of your piston to lift the robot. Also, it’s definitely possible to make a working design without CNC or anything heavy duty, but you’ll probably want to use belts or chains if that’s the case, so you can tension out any discrepancies in hole spacing. We’ve actually already started on a very crude prototype with chains, but we plan on upgrading to gears in the final design.
As for how long it takes to design, I’d say not very long at all, so long as you have a relatively clear picture of how the finished product will look like. And given the many existing designs you can access, that shouldn’t be too hard.