Crazy Crab

[Akash Rastogi]

Originally Posted by antoineft

We attempted a very modular crab drive this year (actually our first time building one). Basically the system is broken up into two almost identical modules that can be removed and replaced in the robot with only 4 bolts per side.

Each side module fits inside a 2" x 6" aluminum tube which keeps it compact and also safe since there are no open running chains. Most of the pocketing seen in the pictures is for getting your fingers in there to assemble it and also to just look nice That being said, they do take patience to assemble! Have some modifications in mind for next time to make assembly easier but overall we were happy with how it turned out.

They also have integrated, fairly simple chain tensioners which serve to also wrap the chain around the drive sprockets in addition to tensioning.

We have a spare module made up to swap in if we have a problem with one of them as well as for display. If you are in Finger Lakes this weekend, or Waterloo stop by to have a closer look. Can't wait to try these babies out in competition!!

Antoine Trabulsi
Lead Mentor
Beaverworx - Team 2609

Wow that is some great stuff! Very nice. I've never actually seen a modular crab drive but that is just awesome and looks very doable. Mind sending over some CAD? Very impressive work. I'm def really impressed with your overall robot and team. Amazing, can't wait to see it in action!

[Galum]

We designed a swerve/crab drive last summer and built it for this year. Our main goals were simplicity and modularity as we do not have a factory/CnC to help, everything is hand built. I have several more efficient and complicated CAD's and designs at home, I'll see if I can dig it up later.

If you are interested, you can view our current system here: http://hamosad1657.net/inventor/Chassis.html

[big1boom]

How are you managing to keep the wheel's vertical if their is only one support?

[antoineft]

Sorry for the delay in getting the CAD up guys. Wanted to wait a bit anyways so I could give you the full scoop after its been battle tested through 2 regionals. Seems a little better than just posting about CAD files.

Overall the system has been amazing and we have loved answering all the questions and showing the drive to everyone who came by to ask about it at both Finger Lakes and Waterloo. Being just a 2nd year team with a brand new driver using crab for the first time, it took a full regional to get the programming tweaked and the driver used to using this drive system. We obviously had a better time driving around for most of the Waterloo regional

We used 1/2" aluminum drive shafts which I am happy to say have held up great (was a little worried to tell the truth). #25 chain for everything. The only real failure we had was one of the tensioner sprockets coming off the bearing that it was press fit onto at FLR. Probably not the best press fit in the first place, but we solved it by drilling 3 small holes into the sprocket to hold the sprocket on with #6 bolts.

I've uploaded the iges file (24MB) for anyone interested here:
http://www.evilshare.com/26a2916e-6f...8-0007e90cfb90
just enter the short code into the box, then hit the "Click to download" button, should be all good.

Any other questions just ask, or feel free to come by in Atlanta as we should have the programming tweaked even more by then. Trying to get some good traction control going as well so should be fun!

[JesseK]

On top of all of the suggestions here, browse CD-Media. It will take you several hours, but you will find drive trains you're looking for. Once you find a good pic, hit the little heart icon so it saves it to your favorites for future reference. More importantly though, you'll have a better idea of what questions to ask.

Another alternative is linkage drive.. It has the benefits of strafing & diagonal movements while also being simpler to implement than most common crab drive trains. This year (2009) my team did that drive train with 4 independent pneumatic pistons for the linkage actuation, and I must say that it was the best decision we've ever made post-kickoff. It was easy to prototype in 2 weeks, and has been the most reliable drive train we've created to date (though I wonder if that has to do with the fact that we have no chains...). I does have its cons, such as the fact that we think our current specific design would explode with high traction wheels on carpet, but there are challenges to overcome in any design if you think about it.

[darthfroggy]

Wow, thanks for all the help. I am noticing that many teams use Bevel gears and many teams are using gear boxes on each wheel pod.
What are the benefits of each and how should i decide which to go with.
Probably the better question is which is easier.

[antoineft]

Originally Posted by darthfroggy

many teams use Bevel gears and many teams are using gear boxes on each wheel pod.
What are the benefits of each and how should i decide which to go with.

I would say it mostly depends on your resources. Its easier to build a crab drive with a gearbox on each wheel pod, but it will also weigh more. If you have the machining capabilites and are up for the challenge go with the bevel gears.

[IceStorm]

I've seen designs for both 3 and 4 wheel drive crab systems. Can any teams out there give advantages/disadvantages to one over the other.

Obviously one of the major advantages to the 3 wheel design is that it would be lighter. But I think it would have the disadvantage of possibly being more tipsy.

We are a 2nd year team and are looking to design a prototype crab/swerve frame during the off season so that we have an idea how to build one if they game calls for one being the better idea in the future. Up until now we have only had tank steered fixed wheel designs with a single gearbox design on per back wheel.

This year we did incorporate some custom analog hall effect sensors so we have had experience with the code stand point of making an encoder work and feel up to all aspects of the design.

Thanks for such a great thread and all the attached posts.

Don't suppose any team would have some basic example labview or C code from the programming stand point that they would be willing to share?

[big1boom]

Originally Posted by antoineft

I would say it mostly depends on your resources. Its easier to build a crab drive with a gearbox on each wheel pod, but it will also weigh more. If you have the machining capabilites and are up for the challenge go with the bevel gears.

It may be easier to do a gearbox per wheel pod. But we were able to pull off bevel gears.

How we machined the wheel pods
1. CAD like crazy
2. Cut all the plates out of 1/8" aluminum, with our 20-year-old wood CNC machine
3. Drill press some axle holes
4. Hand weld the plates together.


Bevel Gear Crab/Swerve
Pros:
Lightweight
Unlimited turning of the wheel pod
Few motors

Cons:
Harder to work with
Less power, however that was not an issue this year.

Gearbox per swerve module
Pros:
More power per wheel
Easier to work with

Cons:
Limited turning because of wiring.
Heavy
Uses up motors that could be used somewhere else


That is all I can think of right now, with limited experience with swerve/crab, I am sure that I forgot something

[Fe_Will]

I was thinking of starting a new thread but I'll just ask this here:

Can people/teams with actual crab/swerve experience post some info on:

what the pros & cons of controlling the 'cans':
individually vs left/right vs front/back vs all together

and/or

what the pros & cons of powering the wheels:
individually vs left/right vs front/back vs all together

Our team is also interested in prototyping this type of drive over the summer, we know the basics but not the advanced control part

(and yes I spent 4 hours going through the pictures/threads that are on CD)

[roboticWanderor]

let me restate that even though Crab can be an amazingly awesome drive train from a mechanical standpoint, it is hell on your coders. design in the proper sensors (absolute encoders, gyro's etc.) and adjustments(chain tension and wheel alignment). otherwise you will never get it to drive the way you want.

BTW: 118 has a great pit display of their coaxial crab modules from like 5 years ago. Have them explain all the in's and outs of their system it's really cool. and if you look closely, it's not just the modules that make the magic happen, it's the whole system, the chain adjustment, gearboxes, sensors, and motor placement that make's crab an elegant solution to drive train mobility.

[Natchez]

Mason and I will be presenting the Robonauts' crab drive at the Championships Conference on Thursday at 8:00 AM. The presentation will include an in-depth look into our evolving crab's mechanical design along with discussions about design trades, lessons learned, and variations. We'll have wheel modules from our past five robots along with Pro-E models available for copying onto flash drives. Also, we're confident that constructive discussions about alternative designs will emerge.

Please come join the fun,
Lucien

[keehun]

Originally Posted by Natchez

Mason and I will be presenting the Robonauts' crab drive at the Championships Conference on Thursday at 8:00 AM. The presentation will include an in-depth look into our evolving crab's mechanical design along with discussions about design trades, lessons learned, and variations. We'll have wheel modules from our past five robots along with Pro-E models available for copying onto flash drives. Also, we're confident that constructive discussions about alternative designs will emerge.

Can some of that information get posted for teams not fortunate enough to go? Like myself?

Thank you,
Keehun

[IceStorm]

That would be awesome if you could make the presentations available online. Our team would greatly appreciate it.

[JesseK]

Originally Posted by Natchez

Mason and I will be presenting the Robonauts' crab drive at the Championships Conference on Thursday at 8:00 AM. The presentation will include an in-depth look into our evolving crab's mechanical design along with discussions about design trades, lessons learned, and variations. We'll have wheel modules from our past five robots along with Pro-E models available for copying onto flash drives. Also, we're confident that constructive discussions about alternative designs will emerge.

Please come join the fun,
Lucien

Is this an official conference? It's not on the current schedule, but it's definitely one I am interested in attending.