Rover: QuickTune velocity tune gains too high

[xianglunkai]

I found that PSC_ The parameters set by VEL are too large, resulting in oscillation behavior after turning at large angles.
My personal opinion is:

1. Set P first instead of D, and D should be relatively small to prevent noise diffusion;

2. I can be set to 0 or 10% of the P value;

3. The maximum setting of D value should be 20% of P value;

[rmackay9]

Thanks for this. I have a few ideas:

1. remove the velocity tune because it doesn't work well enough
2. the Slew Rate calculator can only detect relatively high frequency oscillations (3hz or higher). Rover/Boat oscillations may be lower frequency than this meaning that we need to increase the SlewLimiter.cpp's WINDOW_MS definition to 1000 (or higher). In my simulator tests however this did not help
3. try a different approach to find a good tune. For example we could set the I gain to zero, then increase P until the root-mean-square error reduces.. perhaps then we could continue increasing it until it starts to deteriorate.

[xianglunkai]

@rmackay9
I agree with your suggestion 1. In fact, manual adjustment of PSC is not difficult and the effect is acceptable.
Also, why do we always start setting from D instead of P?

[rmackay9]

@xianglunkai,

Cool, I'll go ahead and remove it then (or you can of course).

I think we do D before P because tuning P may lead to us needing to reduce D.

[rmackay9]

I've created a PR here to greatly simplify the rover quick tune #25629

[rmackay9]

We've removed the velocity tuning from the quick tune process so I'll close this issue.