Hi everyone, I hope anyone who knows a bit about arduplane code can help.
I recently had a crash with a plane which normally flies really well however I added a vision nav system.
I have a simple plane running 4.3.7, matek DLVR pitot, matek m10GPS (index 0, secondery), vision nav talking as a GPS (index 1, primary).
After a few crashes, I found that the ground speed and vector from the GPS were volatile and unrealistic.
The innovations and their square roots of the airspeed were incredibly high as they were compared to the groundspeed vector.
From what I understood from the code if the ‘tasTestRatio’ is above 1 the measurement is rejected.
In the last crash, you could see the AHRS2 airspeed suddenly jump and follow the groundspeed.
At the same time, the TECS true airspeed was stuck and did not change.
(Unfortunately, I am describing it only with words as I lost the logs)
And so I made a few changes since I only care about the location from the vision nav.
1: I read the airspeed data from Mavlink and fed it as the ground speed and as the vector (by calculating the angle of the positions).
2: changed the parameter EK3_EAS_I_GATE from 400 to 1000 which took the ‘tasTestRatio’ down to a good level.
Now the data looks better in the logs and I also can freak the vision nav location (updating it live) without the XKF4->SVT going up.
I have two things I have yet to figure out.
1: when flying one would think the airspeed and ground speed would be equal in this configuration.
However, they are not. What I think is happening is that on takeoff a certain wind estimation vector is formed (which after creation is steady most of the flight) and impacts the ground speed calculation (and the innovation) throughout the flight.
What I found in the code:
wind->sqrtF(sq(stateStruct.velocity.x) + sq(stateStruct.velocity.y)) - tasDataDelayed.tas
‘tasDataDelayed.tas’->defaultAirSpeed * EAS2TAS
defaultAirSpeed → I’m lost here
My main question is how to stop the initial vector formation or the estimation all together.
2: in this case, the change in position makes a certain lower ground speed then the reported ground speed by the vision nav.
I did not find a problem with it however I’m quite scared that this will arise as a problem later on.
log from new test flightlog
In XKF2 you can see the wind vector formation at the start.
I know I have not been quite clear in writing this.
I would be happy to clarify.
thank you all.
nimrod.