Wondering if anyone is willing to tell me what their typical settling time of a swerve drive is just so I can benchmark our PID tuning. Say a step command to rotate 90 degrees. How long does the pod take to rotate to the command angle?
This should take a small fraction of a second. You will not be able to actually time it (although you may be able to plot it to measure it). It will depend on your gearing a bit.
Relative to your PID tuning, what you want to look at is actually the overshoot and the oscillations. Before we tune our PID, we see lightly damped oscillations where the angle will go 10 or 20 degrees past the desired angle, then come back and overshoot the other direction and oscillate back and forth for a few seconds until it settles into the correct angle. The PID tuning is primarily to add enough damping that you see very little overshoot (a couple of degrees is OK but not a large overshoot) and no oscillations. This is generally pretty easy to do and you still end up in a response that hits the desired angle in a small fraction of a second.
We are programmatically measuring it. We are getting a settling time with little to no overshoot around 500ms traversing at about 90 degrees of rotation. We are working on traction/differential swerve drive.
and the gear ratio to control the swiveling is far less than other traditional swerve designs. It’s about 8:1. But we are using 2 NEO’s to power the rotation so in the end the torque available is similar.
Our mass momentum of inertia is a bit larger since we are rotating the motors and gearing, about 0.0184 kg-m^2.
It seems to be handling well. I’m just trying to benchmark our design against a traditional swerve. I’ve seen videos of traditional swerves swiveling and settling incredibly quickly, on the order of 100ms is what I would guessed. Just wondering if that is unusually fast or typical.
I don’t have data handy to look at, but I would saw that 100 ms is a lot closer than 500 ms for 90 degree rotation.
Having said that, 500 ms may not be a problem for driving. It does seem a little slow, but I’m not sure how it would behave in real driving situations.
This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.