Enhancement: More calibration related parameters are preserved when o…

[limhyon]

Describe problem solved by this pull request
Current rcS script invalidated magnetometer calibration data when SYS_AUTOCONFIG sets to 1 with AUTOSTART number given.

Describe your solution
More calibration related parameters are preserved when only autoconfig reset happened. Current setup makes magnetometer and offset data invalidated.

[dagar]

I was opposed to some of this in the past, but I think it's worth reconsidering. We have preflight checks for the estimator and sensor consistency checks. If there's an issue it's quite likely the user is going to re-calibrate at that point.

If anything I've noticed that unnecessarily resetting calibrations leads to users redoing the calibration quickly and poorly.

Does anyone else have an opinion here? @bresch @bkueng @jlecoeur

[bresch]

Personally, I think that only the autopilot-related parameter should be preserved, so not the bias of the external mag and not the EKF biases.
I know that it is annoying when the user knows that nothing changed and doesn't want to redo the mag calibration, but I think we should add a new SYS_AUTOCONFIG mode that preserve all the calibrations instead of adding more and more to the "reset parameters"

[jlecoeur]

I am generally in favor of this. See #13276.
However I think that the current default makes sense: different airframe means different magnetic bias.

We may discriminate a software airframe change (no calib needed) from a hardware airframe change (calib needed).

Ideally, the user would be prompted just after selecting a new airframe:
"New airframe config selected. Has the autopilot been physically moved on the airframe or mounted on a new airframe?
[.] Yes. Perform magnetometer calibration at next startup.
[.] No.
"
Internally we can map this to SYS_AUTOCONFIG=1 for hardware airframe change (current behaviour) and SYS_AUTOCONFIG=2 for software-only airframe change.

[dagar]

I think we should add a new SYS_AUTOCONFIG mode that preserve all the calibrations instead of adding more and more to the "reset parameters"

Actually, one of the cases is SYS_AUTOSTART not changing (re-clicking the same airframe in QGC). It would be safe to preserve mag cal in that case.

@bresch that could work longer term, but I think in most cases it's really the QGC side driving this.

@hyonlim could you go through your particular usage? Would the special case around an unchanged SYS_AUTOSTART help?

[limhyon]

If there is special case for SYS_AUTOSTART, that would help.
When airframe parameter is updated by software, the calibration is invalidated.
I think @jlecoeur 's statement is good.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. Thank you for your contributions.