[EKF2] Run simplified GNSS checks after initial fix

[bresch]

requires #24908

Solved Problem

Thresholds of GNSS checks are supposed to be tight enough to make sure the drone starts with a good initial fix. However, currently, degradation of a single metric also stops the fusion (e.g.: high DOP) when other metrics are still good.

Solution

Add a "simplified" list of checks with large (arbitrary) thresholds. Those values can be discussed, but hopefully we can agree on a common value to avoid additional parameters.

Changelog Entry

For release notes:

Improvement: Simplified GNSS checks after initial fix 

Test coverage

unit tests
@haumarco flight tested

[github-actions]

🔎 FLASH Analysis

px4_fmu-v5x [Total VM Diff: 216 byte (0.01 %)]

    FILE SIZE        VM SIZE    
--------------  -------------- 
+0.0%    +216  +0.0%    +216    .text
  [NEW]    +212  [NEW]    +212    estimator::GnssChecks::runSimplifiedChecks()
  +0.0%      +8  +0.0%      +8    [section .text]
  -0.7%      -4  -0.7%      -4    Ekf::updateGnssVel()
+0.0%     +55  [ = ]       0    .debug_abbrev
+0.0%      +8  [ = ]       0    .debug_aranges
+0.0%     +16  [ = ]       0    .debug_frame
+0.0%    +331  [ = ]       0    .debug_info
+0.0%    +294  [ = ]       0    .debug_line
  +300%      +3  [ = ]       0    [Unmapped]
  +0.0%    +291  [ = ]       0    [section .debug_line]
+0.0%    +315  [ = ]       0    .debug_loclists
+0.0%      +7  [ = ]       0    .debug_rnglists
+0.0%     +86  [ = ]       0    .debug_str
-0.8%      -2  [ = ]       0    .shstrtab
+0.0%     +66  [ = ]       0    .strtab
  [NEW]     +66  [ = ]       0    estimator::GnssChecks::runSimplifiedChecks()
+0.0%     +48  [ = ]       0    .symtab
 -33.3%     -16  [ = ]       0    ___ZNK3Ekf21updateAidSourceStatusI24estimator_aid_source3d_sN6matrix7Vector3IfEES4_EEvRT_RKyRKT1_RKT0_SE_SE_f.isra.0_veneer
 -25.0%     -16  [ = ]       0    __nxsem_tickwait_uninterruptible_veneer
   +50%     +16  [ = ]       0    __stm32_i2c_set_bytes_to_transfer_veneer
   +33%     +16  [ = ]       0    __stm32_recvfifo_veneer
  [NEW]     +48  [ = ]       0    estimator::GnssChecks::runSimplifiedChecks()
-2.0%    -216  [ = ]       0    [Unmapped]
+0.0% +1.20Ki  +0.0%    +216    TOTAL

px4_fmu-v6x [Total VM Diff: 216 byte (0.01 %)]

    FILE SIZE        VM SIZE    
--------------  -------------- 
+0.0%    +216  +0.0%    +216    .text
  [NEW]    +212  [NEW]    +212    estimator::GnssChecks::runSimplifiedChecks()
  +0.0%      +8  +0.0%      +8    [section .text]
  -0.7%      -4  -0.7%      -4    Ekf::updateGnssVel()
+0.0%     +55  [ = ]       0    .debug_abbrev
+0.0%      +8  [ = ]       0    .debug_aranges
+0.0%     +16  [ = ]       0    .debug_frame
+0.0%    +331  [ = ]       0    .debug_info
+0.0%    +286  [ = ]       0    .debug_line
 -71.4%      -5  [ = ]       0    [Unmapped]
  +0.0%    +291  [ = ]       0    [section .debug_line]
+0.0%    +230  [ = ]       0    .debug_loclists
+0.0%      +8  [ = ]       0    .debug_rnglists
  +100%      +1  [ = ]       0    [Unmapped]
  +0.0%      +7  [ = ]       0    [section .debug_rnglists]
+0.0%     +86  [ = ]       0    .debug_str
-0.8%      -2  [ = ]       0    .shstrtab
+0.0%     +66  [ = ]       0    .strtab
  [NEW]     +66  [ = ]       0    estimator::GnssChecks::runSimplifiedChecks()
+0.0%     +48  [ = ]       0    .symtab
  [NEW]     +48  [ = ]       0    estimator::GnssChecks::runSimplifiedChecks()
-4.4%    -216  [ = ]       0    [Unmapped]
+0.0% +1.11Ki  +0.0%    +216    TOTAL

Updated: 2025-06-05T09:14:59

[haumarco]

Why do we want to always use the simplified checks? Shouldn't this be something which can be configured by the user? Are the current thresholds not really, really high? Or should it simply be a better sanity check and we let the test-ratio do the proper evaluation?

[haumarco]

After further discussion and comparison to previous versions of GNSS checks in PX4, I also think this is a good approach to loosen the checks. About the specific magnitude of the new thresholds I'm still not 100% sure since I could not test with "real"/in-air scenarios where GNSS gets bad all of the sudden.

Here's a test of starting outdoors and then moving indoors.
log_142_2025-6-6-15-19-52.zip

[DronecodeBot]

This pull request has been mentioned on Discussion Forum for PX4, Pixhawk, QGroundControl, MAVSDK, MAVLink. There might be relevant details there:

https://discuss.px4.io/t/wiki-errors-for-ek2-gps-check/45940/10