Second-Order Swerve vs? Discretized ChassisSpeeds

[narmstro2020]

So this post on CD

brought up two key ideas, but I never quite got this jist on whether Discretized Chassis Speeds was equivalent to Second Order Swerve or a separate issue and on whether or not you could implement both Second Order Swerve and also Discretize chassis Speed for added reduction in Swerve Drive Skewing.

[calcmogul]

You should use one or the other, not both. ChassisSpeeds.discretize() is WPILib's solution for the skew issue caused by sample delay. Remaining skew after that compensation is from poor feedforward/feedback control, in my experience.

[narmstro2020]

Awesome. Thanks

This year we found using the discretize() method worked better for us but only after setting the time to something in the 80's of milliseconds for the time argument.

I'm guessing ideally this should be closer to the control frequency (20 ms if using the default)? We ran everything on the RIO at 20, so I'm guessing poor feedback/feedforward control was in place and we just compensated until it looked smooth enough. Our Kraken drive wheel's and NEO steers were running at there defaults (something in the 1kHZ range I think).

[calcmogul]

Yea, the discretization period should be the same as the velocity setpoint update period.