I am currently trying to design a Octomecanum drive train for fun and possibly as a drive train for my team to use in the future and I have come to a point where I have gotten a little confused. Every example uses pneumatic cylinders to shift from mecanum to tank and I was wondering how people prevent the pistons from bending due to the stress. If anyone who has experience with this or some insight that would be really helpful.
The rod of a pneumatic cylinder is more likely to bend if it experiences a side load or moment at the end of the rod while it is in its extended state. In the retracted state, the rod of the cylinder will be less likely to bend. I do not have specific experience with articulating drives, so maybe someone who does could chime in, but this might be important to consider if you are worried about bending the rod of the pneumatic cylinder. Think about which configuration the cylinder is most likely to experience side/bending loads, then make that the retracted position.
Are you referring to bending that would be caused by robot-on-robot impacts?
I’ve never built one, but in the designs that I’ve seen, the pneumatic cylinder would be allowed to rotate at the blind end, so that as it extends, the cylinder will rotate so that the force always goes along it’s axis.
I am more concerned how people support the cylinder when it is connected to the shifting parts. What I have seen is that there is no support but I wanted to find out if that was true or if I was missing something.
Two questions:
Which method of shifting are you using?
A vertical-cylinder system like this: http://www.chiefdelphi.com/media/photos/37426
Or a horizontal-cylinder system like this: http://www.chiefdelphi.com/media/photos/37096
And given that choice, what kind of “support” are you talking about? Supporting the rod? supporting the blind end? supporting the module?
We have used an octocanum for the past year. We used no supports on the cylinders. We have had no known bending of the shafts themselves.
Optimally your piston should be a simple two force member, that is, only experiencing direction extension or compression loads. Many times this is accomplished via ball joint fittings, or having enough play in its pivot that it doesn’t take any load.
Keep check on load paths through the module and it should be easy enough to be sure the piston isn’t taking any bending loads.
-Aren
To expand on Aren’s points, if the cylinders are loaded as he described the rod size is spec’d such that you won’t ever fail a cylinder from it’s own force.
A somewhat related question: Does 148 (or neutrino) implement a encoder into their butterfly drive?
Neutrino ran last year with no sensors on the drivetrain, but still had a 7 disc auto 
Did they just do gyro PID and then drive for set times?
I doubt they used gyros. Their butterfly allowed them to drive straight without PID.