Hi, I want to teach Robot Characterization to my team but half of them have Macs. It appears that the SysID tool is not available for MacOS and the previous Robot Characterization Tool has been discontinued. Are there any plans to port to Mac? I couldn’t find a release for Mac on Github.
Thank you.
SysId 2022 supports x86 macOS but not M1 ARM macOS. I assume your team has a bunch of M1 macs. SysId 2023 will also support M1 ARM.
You won’t find a release on the SysId repo because the tool is distributed with the WPILib installer. There’s an x86 macOS installer listed here: Release WPILib 2022.4.1 Release · wpilibsuite/allwpilib · GitHub
In general, x86 macOS stuff tends to work on M1.
Also when will SysID get a renaming to match what it actually does. It’s a great tool, bit it sounds like a windows background process.
(Mostly /s, but not fully /s)
It’s a system identification tool. It’s named after exactly what it does.
Better named: ChampsUnlocker3000
Thank you. Sorry, I had the 2022.2.1 version installed. I see SysID under the Start Tool now.
Most of the team have the Intel Macs so we’re probably good to go. I got it launched OK on my M1 from the command line so I’ll try it out. Thanks again!
It’s a system characterization tool right? That’s different than identity
From the Wikipedia page:
The field of system identification uses statistical methods to build mathematical models of dynamical systems from measured data.
This is exactly what SysId is doing under the hood. Determining the elements in the matrices of a linear model of known dimensionality is still identification.
Also, all the literature uses the phrase “system identification”, not “system characterization”.
Any “field” which claims it has a monopoly on the term “system identification” needs to leave their lab, walk outside, and give their head a shake. It’s a completely unclear, unteachable and ungooglable term, and I’m sure I’m not the only one to think so.
(FWIW, every time I’ve used a similar system at work, it’s been called a “PID Tuning Tool”)
/shakes fist at clouds.
You’re identifying the system/model that, for a set of measured inputs, makes the measured outputs match the estimated/expected outputs. That’s where the name comes from. In other words, it’s a model selection process.
Turns out Google agrees with you and not me. Now I’m in the unenviable position of thinking Google is wrong too.
(The big secret is that our model selection process, at the moment, is still just a linear regression…)
Coming this spring: FRC’s Next Top Model
Ah, memories of the school network blocking certain parts storefronts/sites because they contained the word “model.”
This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.