Weird lateral ramp logger plot?

[ruiyuansd]

Robot Setup: Mecanum Drive with Three Deadwheeel Localizer
Beacuse of the difference between (kS_forward, kV_forward) and (kS_lateral, kV_lateral), the ramp logger regression plot is not ideal. Any suggestion on that?

[rbrott]

This plot exposes a limitation of the robot model. The additional friction dynamics when strafing cannot be captured by a single multiplier but require an additional offset at least. On the bright side, this was just as broken in previous versions of Road Runner, and feedback can make up some of the difference.

[pn89348]

In this case, is it better to use the regression line similar to what's shown in the screenshot or to exclude the flat portion at the beginning?

[rbrott]

Theoretically you should remove the flat portion, though I don't expect it to make a big difference in the overall fit because the line must still pass through the origin.

[ClarkBrendan]

I think we're wrapping too much into lateralInPerTick. The name suggests that it is a direct ratio between the inches and ticks of the encoder (either deadwheels or drive encoders). However, that's not how it's being used here and that makes the code significantly less readable.

Instead I suggest we use lateralInPerTick in the way the name suggests (defaulting to inPerTick when using deadwheels), and 2 new variables that are tuned by lateralRampLogger, lateralKs, and lateralKv.

Thoughts?