AP_Baro: Align processing procedures #25199
Conversation
There was a problem hiding this comment.
That isn't the same.
On first one, we are just discarding the value over the 1 second. So we don't care at what rate we get them
On second case, we want an average of 5 values. Best woould be if they can be continus, in worst case senario the change would lead the everaging to be slightly off. The behavior is already present but limited by the fact we don't add 10ms delay between reading and just try to get the data as they come.
Using a median can help to have better sampling and discard extrem values
|
As @khancyr notes, this is a significant change to the way we sample the baro for the calibration, and the two steps you're trying to make uniform are fulfilling different functions. We'd need a pretty good reason to change this up, as the existing code has worked well for a very, very long time. It's also currently objectively worse as it takes longer to calibrate the baro! I'm going to close this one now - but if you can show that this is a signifcant advantage for some reason, please reopen and explain. |
Same pre-processing.
I think a delay time is needed.