2 Wheel Swerve (2 swerve, 2 omni)

[ENIAC]

I had an idea for a novel drivetrain (or at least what I think is novel, someone may already have come up with this), and I wanted to have it checked out:

With a diamond wheel formation, the two corners on the long axis have swerve wheels mounted, and the corners on the short axis have omni wheels mounted, with their plane of rotation parallel to the long axis. The two swerve wheels are chained or belted to allow for simultaneous turning.

Basically, it allows for omnidirectional motion, but uses fewer swerve wheels and so is less mechanically complex*. The downside is that traction falls off towards one direction, but that's a sacrifice you already make with tank. Allows for most of the capabilities of swerve as far as I can tell.

Thoughts?

*call me out on this if I'm wrong

[Alan Anderson]

It seems like a reasonable approach, but the diamond configuration seems unnecessarily tippy. I think a traditional corners-of-the-rectangle wheel placement should work as well. It shouldn't even matter which two corners the two steerable wheels occupy.

[DampRobot]

I believe team 100 tried this approach in 2005, although I could be thinking of another team. Instead of a diamond pattern, they had 6 wheels (with the swerve modules in being the middle two of a two by three grid of wheels). I believe they had problems keeping the modules rotating together, which didn't result in a very effective system.

If you're simply looking to eliminate weight, check out some of Bomb Squad's past drivetrains. In addition to an awesome, iterated into oblivion swerve set up, they've tried 3-wheeled swerve. As far as I can remember, it had great performance and substantial weight savings.

[Tom Line]

One of the most famous bot's in CD's history had a very similar drive:

http://www.chiefdelphi.com/media/photos/12948

[Lil' Lavery]

See Wildstang and Simbotics (ball casters as opposed to omni-wheels) in 2004. The Simbots were even kind enough to post this whitepaper on their drive:
http://www.chiefdelphi.com/media/papers/1552

[ajlapp]

We did one back in college...

http://www.chiefdelphi.com/media/photos/14645

[Brandon Zalinsky]

Are the two omni wheels powered?

[Christopher149]

Originally Posted by ajlapp

We did one back in college...

http://www.chiefdelphi.com/media/photos/14645

Hello from Michigan Tech. Part of the lower arm of that robot (with the Michigan Tech cut-out) is currently hanging on our cage in the high school where we store the robot. I didn't realize that we had ever done any swerve on the team.

[ajlapp]

That was the only version we tried...it was reasonably successful. We had individual steering using a globe motor. The transmission was integrated into the module.

On paper it was a really great bot.

http://www.chiefdelphi.com/media/photos/14647

[ENIAC]

Originally Posted by Flak-Bait

Are the two omni wheels powered?

Yes, they are.

Originally Posted by Alan Anderson

It seems like a reasonable approach, but the diamond configuration seems unnecessarily tippy. I think a traditional corners-of-the-rectangle wheel placement should work as well. It shouldn't even matter which two corners the two steerable wheels occupy.

I was thinking that as well. I saw the example in Post #6 with the omnis and the swerve modules on separate sides.

Would there be an optimal driving orientation, assuming the omnis are at opposite corners? i.e, would it drive straight better when the sides are orthogonal or parallel to the direction of movement, or when they are at 45 degree angles?

Sorry if I am being unclear.