fix(ekf2): apply baro ground effect deadzone symmetrically#26655
fix(ekf2): apply baro ground effect deadzone symmetrically#26655dakejahl wants to merge 2 commits into
Conversation
The baro ground effect deadzone was only applied to negative vertical position innovations, assuming ground effect always causes a positive static pressure transient. In practice, the direction of the pressure transient depends on the baro location on the airframe, so ground effect can produce large innovations in either direction. Apply the deadzone symmetrically to both positive and negative baro innovations.
|
Do you have a log showing this? |
Yes, this is my 3" FPV flying inside |
|
Sampled 20 recent multirotor logs from the public flight review database and checked Result: 15/20 (75%) showed positive-dominant baro innovations, 2 negative-dominant, 2 mixed, 1 negligible. The current one-sided deadzone does not mitigate the majority case. Five representative logs with clear positive innovations during takeoff:
|
|
I have a plan to tackle this better -- with and without a rangefinder. We need baro thrust compensation first though #26924 |
2 participants
The baro ground effect deadzone was only applied to negative vertical position innovations, assuming ground effect always causes a positive static pressure transient (pressure rise). In practice, the direction of the pressure transient depends on the baro placement on the airframe, so ground effect can produce large baro innovations in either direction.
This applies the deadzone symmetrically to both positive and negative baro innovations, and updates the related comments, param descriptions, and control status flag description to remove the one-directional assumption.