The one thing I can’t stop thinking about is the solution to achieving omnidrive without slipping around when pushed.
My team did kiwi drive this year, which worked, and over the past month I thought of the possibility of reorienting all wheels so they spun away from the center of the bot.
Control would be a bit awkward, but after looking at my coding, I found out that kiwi drive code seems to work with some tweaking. That said, the weird part would be that strafing is achieved through spinning the left and right wheels in alternate directions, top wheel as deadweight.
However, if another triple set of wheels was setup so each original wheel had an alternate on the other side, that would solve the problem (“strafing” now being achieved similar to conventional 4 wheel omni). Top view looks a lot like the Walmart logo, hence me calling it “Walmart Drive”. The original intent was in no way geared toward getting donations from Walmart . Since there’s usually a 4 CIM limit, though, this would require a pillar of 3 gearboxes in the middle powering each wheel pair.
The difference I’m thinking it has over kiwi or most omnidrives is that now the bot is fairly difficult to push as the wheels spin outward instead of the rollers rolling outward. I believe it still works as an omnidrive in this configuration, though I’m not totally sure. Nor am I sure whether it’s worth doing in build season. Still, it seems like something fun to play with
This configuration will work fine as an omnidrive, but removes your ability to turn.
And allows others to spin you like a top.
If this is the one feature that you want in an omnidrive, try swerve. It provides something very close to true omnidrive while providing a lot of traction (we won a few pushing matches against tank bots in 2011 with our swerve). Other than swerve, I don’t really see any quick and easy solutions to a “non pushable” omnidirectional drivetrain.
Team 221 sells a good kit for swerve, albeit the modules are heavy.
Does anyone else have other ideas?
You still lose a large portion of your available traction compared to a system in which all wheels work in the same direction (usually either a skid-steer or swerve). Any time you have something supporting the weight of your robot that does not contribute to the forward motion, or does not contribute 100%, you end up with less maximum traction (which is what often decides pushing matches, and is a major reason why many teams avoid omnidrives).
Its not even much better in that respect than a “conventional” omnidrive (with 4 omniwheels oriented 90 degrees to each other):
Depending on which direction you are being pushed in, a 4 wheel omnidrive has between 50% and 71% of the maximum available traction. The “Walmart” drive would give between 58% and 67%.
It does look cool, though.
I think it is an interesting idea for an off season project for you team. Try it out and see if you think is correct. I’d love to see how and if this design could work. It does look like the programmers would have several sleepless nights trying to get it to work.
The fact that you cannot turn yet others can turn you easily negates any advantage of flipping the wheels. Plus, driving each wheel is going to be hard. Though using normal orientation wheels but 6 of them is an interesting idea.