many differential swerve ideas I’ve seen are one rig gear on top and one on the bottom with a gear in the middle acting as an axle
This design that I’m thinking about prototyping is essentially two wheels on a single axle that are both powered so that when one is getting more power than the other it causes the axle to turn the main idea from there would be that the whole assembly would be on a bearing so it could account for the shifting.
Feel free to critique this as much as you want theres probably something I’m missing as to why this wouldn’t work and if not feel free to prototype this for yourselves.
One issue you will run into is wire routing. With a normal swerve/differential swerve you can spin the swerve module as many times as you want. With your design you will be limited in rotation due to the wires. That is unless you find a way to use those slip rings that transfer power. Would love to see a design though I hope you can get something working
that is something i was thinking about getting around I was between the idea of having a limited rang of movement since in practice you probably only really need 180 degrees of movement or having a lot of extra wire length and a harness to hold it in place
Yea you could totally code in a 180 only rotation and have the swerve swap directions after it hit that 180. Would probably be a pain to code but im also not a programmer so.
That happens to make the swerve nearly undrivable. Flipping the module by 180 degrees uncommanded by the driver causes a jerk in heading that makes driving the robot really annoying. (Trust me, I didn’t realize how terrible this is until we fixed the software issue that caused it)
Also, controlling a module’s steering would become next to impossible with this design. Imagine if one or both wheels are off the ground, then whatever control loop you have running on the steering axis would experience serious integral windup as the motors spun without causing a change in the actual steering angle.
Finally, you wouldn’t be able to use the Falcons’ integrated encoders to get the steering angle because any amount of slippage between the wheels and the carpet would cause an unacceptable loss of precision. (TL;DR: You’d need a separate encoder on the steering axis for this to work)
What I meant by this was swap the drive direction not the whole module. So in simply terms when it hits the rotation limit it goes from forward into reverse and continues. Still not sure if it would work well or not though.
We created a similar designed and abandoned the idea. We added slip rings and suspension to account for all the problems stated above. The issue is gear ratio. In order to have the correct steering and drive gear ratio you either need to spread your wheels apart by at least 4" and have a wheel diameter less than 2.5". But packaging and weight made this undesirable.
That was our first differential swerve built in 2017. Pretty much as soon as it first moved on carpet, we knew it was not going to be a practical design.
I think we are on our 6th differential swerve chassis. The 6th one is this season’s robot. It is going to Houston and most people don’t even realize its not a traditional swerve. They should all be documented here. Search out our team’s posts before you go down that rabbit hole.
