The section on sysid details kS, kV, and kA... but the differential drive simulator only deals with kV and kA and completely ignores kS. I found that the behavior in simulation was close, but still noticably different from Real Life until I accounted for kS.
The dumb-simple way I integrated kS (linear only) was with the following:
I haven't done anything rigorous yet like I did for the drivetrain K_v/K_a sysid model, but the threshold is probably some kind of weighted average between K_s,linear if V_l = V_r and K_s,angular if V_l = -V_r.
The section on sysid details kS, kV, and kA... but the differential drive simulator only deals with kV and kA and completely ignores kS. I found that the behavior in simulation was close, but still noticably different from Real Life until I accounted for kS.
The dumb-simple way I integrated kS (linear only) was with the following:
Once I added that.. the robot movement was much closer to what I expected. I suspect it's not quite right though.
It's possible that one result of this documentation request might end up as an addition to the simulation classes in addition to the documentation.