ekf2 preflight: only check innovation of active height sources #19774
Conversation
There was a problem hiding this comment.
I don't see any error. This follows the same pattern as other innovation checks. I just had the question in the comments about the filter updates.
I quickly tried on my setup and the "height estimate unstable" blocking error is resolved. It keeps warning me that I don't have a GPS fix but as discussed this is actually useful and not blocking flight.
| _preflt_checker.setUsingBaroHgtAiding(_ekf.control_status_flags().baro_hgt); | ||
| _preflt_checker.setUsingGpsHgtAiding(_ekf.control_status_flags().gps_hgt); | ||
| _preflt_checker.setUsingRngHgtAiding(_ekf.control_status_flags().rng_hgt); | ||
| _preflt_checker.setUsingEvHgtAiding(_ekf.control_status_flags().ev_hgt); |
There was a problem hiding this comment.
Note: I had to look up what Ev means again. Maybe the naming could be made more readable 😅
| bool has_failed = false; | ||
|
|
||
| if (_is_using_baro_hgt_aiding) { | ||
| const float baro_hgt_innov_lpf = _filter_baro_hgt_innov.update(innov.baro_vpos, alpha, _hgt_innov_spike_lim); |
There was a problem hiding this comment.
Shouldn't the innovation low-pass filter always be updated and reset if it's not updated?
As far as I can see it's only reset up on landing but as we found out with this bug accepting the altitude estimate (and allowing Altitude mode) is dependent on the result of this check. I assume also in air.
|
We should come back to this later and find a way to do it without manually mirroring so much state. |
|
Im able to arm as well but I do get GPS Disabled in QGC Daily. No other changes to the code (custom airframes only on this test). EDIT: I should note that when it goes back to not having a stable enough GPS (preflight fail, etc) it says AHRS disabled.
|
|
Did a couple of flights. Still had the GPS disabled message but after GPS had enough sats and what not, it didn't give the fail message (but still the gps disabled stayed). Strangely no logging was recorded even though it says the log was started (see attached). Have the tlog but that's all. Not sure if a one off but did several flights. All the same |
|
Flight was fine in position mode, but just like Ryan. GPS shows "disabled" in QGC sensor status. https://review.px4.io/plot_app?log=5cc286d1-91b7-4662-abcd-f623b50037d8 |
fixes #19770
Describe problem solved by this pull request
In 639222d#diff-b98e0cba6406890a752c8c8deb36086c8ed27442210a00a7ac02e2692245460c , the innovation for all the potential aiding sources is published. The problem is that the preflight checker in EKF2 module assumed that only the active source was non-zero (which was correct before this commit).
Describe your solution
Only run the check on the active height source(s), similar to what we were already doing for the horizontal sources.