Quadplane: Transition fails if blowing with the wind during VTOL climb

[AndKe]

Bug report

When starting in a significant headwind, it may be tempting to drift off backward, in order to do the transition closer to oneself. / or to save energy, rather han holding position during VTOL climb (pitching nose down towards headwind)

To reproduce: 1 Climb in Qloiter or Qstabilize, keep nose into the wind, let the UAV drift "backward" 2 when at altitude - switch to (for example) Cruise 3 observe that the transition fails.

To me, it seems that the wind estimate goes bad when drifting/flying backward during VTOL climb.

Version AP4.4.0 (and newer)

Platform [ ] All [ ] AntennaTracker [ ] Copter [ x] Plane [ ] Rover [ ] Submarine

Airframe type Quadplane

Hardware type Cube orange

Logs https://drive.google.com/file/d/1LPinfPQbLsfKwo_GzhEuuNPsuV78kXQ3/view?usp=sharing

[IamPete1]

In this case you have a airspeed sensor so the wind estimate is not being used.

It should not update when the VTOL motors are active. However we do get one change when you first switch into cruise:

@tridge I think this is a race between in_vtol_mode going false and in_assisted_flight going true in update_fly_forward. The delay in changing from false to true we were talking about should fix this too.

In this case it looks like cruise locked in the flying backwards ground course, so it tried to do a 180 deg turn, this is what I get in the sim:

So the fix is just that we are flying forward over the ground before letting cruise lock in heading.

[IamPete1]

@AndKe can you check in FBWB to confirm if it is the attempt to do the 180 turn in cruise that is causing your issue.

[AndKe]

I did not mean to send one of the logs where any significant yaw/was involved, I did not mean to draw the focus to the yaw/heading behaviour, but the airspeed / transition related part.

@IamPete1 It may be some time till I get to fly those UAVs again, it's winter in northern Norway now. I doubt there will be more flying this year.