Team 3863 is proud to present our latest offseason creation: a fully fledged swerve drive! We’ve been looking to make this a reality for some time now, and after blood, sweat, and tears, we’ve managed to pull it off.
~17fps free speed (5.33:1 CIM reduction on 4" wheels)
~5" width x ~6" length x ~10" height (for modules)
~6.3lbs
24" x 24" chassis
Code for shortest rotation arc (turn to complement and reverse motor power)
We had a lot of fun designing, testing, and writing code for this little bot. If you guys have any questions/comments, our team would be more than happy to answer them!
Our answer is most likely no. Although these swerve modules were a GREAT way to verify our new build processes and team management skills, at the end of the day we’re spending 4x the time on the drivetrain when we could be doing other better things in build season. The module would also need a few more tweaks to make it better for rapid manufacture. It’s not completely out of the question though – perhaps when the bag goes away in 2020!
As a long time fan of not spending any more effort than needed on our drive train, boyyy does this make me want to try swerve.
With a sudden rise in crazy swerve designs its nice to see something simple and elegant. I love how you dont need a CAD split view to see whats going on.
Now only if you could find a way to shrink that steering motor, this thing would be tiny with a neo on top.
Swerve is fun. If there was anybody who could pull off a better swerve, it would be you!
I’ve been thinking about ways to shrink that steering motor, since it really bothers me how far it sticks past the CIM. Honestly, it just seems like the best solution is using a 775pro or a BAG and use a 100:12 32DP first stage and something like a 72:12 20DP second stage, and limiting the voltage going into the motors. We don’t need all that power on our steering anyways, and the cost savings from not using a VP really help justify this approach.
I usually appreciate Anand and his designs… but not this one. I don’t love making vertical plates, and at the end of the day, it just seems easier to borrow some ideas from 1323, make them better, and call it good. Who knows, in the end I could be wrong!
The vertical plates are actually just waterjetted and tapped on the bottom (although you could use some jesus nuts instead). The more dangerous flaw is the fact that only 4 tiny bearings protect the module from moment loads when the robot gets hit from the side.
