Autolaunch issue with crash

[Fabio-FPV]

Hello everyone,
I would like to kindly ask for your help with a problem that's stranding me on the ground regarding the autolaunch of my flying wing.
I've been flying with INAV for a long time, and the autolaunch feature worked perfectly in both versions 5 and 6. Since version 7, I've noticed that during takeoff, the aircraft's behavior was to always move to the left compared to the straight trajectory, a noticeable but not impacting its ability to lift the aircraft to altitude.
I've now built another flying wing with the exact same hardware, and in detail:

ARWING PRO
FC SpeedeBee F405 WING
INAV 9

After downloading the Diff_All from the working aircraft, I loaded it onto the new one and recalibrated the accelerometer.
As you can see from the three attached videos, I've written the description below:

Video 1: Maiden Flight, which almost succeeded with a crash
Video 2: Second test with a crash but no log
Video 3: Third test crashed but with Log (attached), but I tried to launch the plane after it reached maximum speed in my hands. I thought it was the torque (between "launch_idle_thr" and "launch_thr") that was causing the roll axis to shift to the left, but I've seen evidence that this isn't the case.

I tried to interpret the log, and it seems to be INAV, which, immediately after launch, even if the plane's attitude was correct, was commanding the servos to the left for some reason unknown to me. You can also notice this behavior by carefully watching the third video.

Now, if I don't have a valid explanation, I won't be able to use it anymore.

Thank you in advance for any invaluable help you may have.

https://youtu.be/dhB_S2aJoh8
https://youtu.be/dP0h9veVK0k
https://youtu.be/k3GP0at08ns

LOG00013.TXT
ARWING PRO INAV 9 Diff All 03_05_2026.txt

[breadoven]

From a quick check of the log PID sum roll would appear to be demanding right roll so I assume the control demand is correct. However, it would appear that the wing just doesn't have enough speed to fly properly. It's already banked over at 60 degs with a speed < 40 km/h which I doubt is enough for the elevons to work effectively. It's also only using 60% throttle which doesn't help with accelerating to flying speed ASAP.

[Fabio-FPV]

Hi Breadoven,
first of all, thank you so much for your kind reply to my issue and for taking it into consideration.
I think you're right, especially since I tried installing Ardupilot software on the same build without changing anything mechanically, and the result was the same (crashes twice).
So, when I was finally about to give up, I finally found that the left elevator was deformed in such a way that even though it appeared to be mirrored, it tended to roll the plane to the left as soon as it gained a bit of lift. I better aligned the length of the left servo's mechanics, compensating for the deformation, and now I'm able to take off.
I wasn't very good at understanding the log, and I thank you for doing so for me.
The fact remains that, as I wrote previously, since version 7 of INAV, autolaunch has gotten worse because it still tends to roll to the left, even on my original build.
Then, in my opinion, the INAV system has enormous potential, and I'm very sorry I haven't been able to find a solution to my other major problem, issue ID 11223 "Wobbles behavior roll axis fixed wing," which unfortunately forced me to temporarily abandon INAV until I can build the Mini Talon Pro again. I think the aforementioned issue ID 11223 deserves further investigation by the developers, because in my opinion, in some circumstances it causes the system to resonance, which is beyond the scope of parameterization, and this is a bug for me.
Thank you again for your invaluable help and all INAV community.

[breadoven]

The fact remains that, as I wrote previously, since version 7 of INAV, autolaunch has gotten worse because it still tends to roll to the left, even on my original build.

I have noticed that even on HITL (simulator flight) there is a left roll when using Auto launch. I always assumed it was just some unrealistic simulation of motor torque effects so ignored it. Checking HITL logs though it does seem the control response is correct so it's trying to correct the roll but can't probably because airspeed is too low at the start of the launch for the controls to work effectively.

Then, in my opinion, the INAV system has enormous potential, and I'm very sorry I haven't been able to find a solution to my other major problem, issue ID 11223 "Wobbles behavior roll axis fixed wing," which unfortunately forced me to temporarily abandon INAV until I can build the Mini Talon Pro again. I think the aforementioned issue ID 11223 deserves further investigation by the developers, because in my opinion, in some circumstances it causes the system to resonance, which is beyond the scope of parameterization, and this is a bug for me. Thank you again for your invaluable help and all INAV community.

Did you try the same wobbly wing with Ardupilot ? If so was the roll oscillation still present ? I've found flying wings to be inherently "wobbly" compared to conventional aircraft although it tends to be more an issue in pitch than roll perhaps. The oscillation on that wing looks a lot like Dutch roll which can be a problem with these wings as I understand it. Did you try physical tweaks to the model, e.g. move C.G., servo throws etc ? Also increasing cruise speed would help since wobble gets worse the slower you go.

[Fabio-FPV]

The fact remains that, as I wrote previously, since version 7 of INAV, autolaunch has gotten worse because it still tends to roll to the left, even on my original build.

I have noticed that even on HITL (simulator flight) there is a left roll when using Auto launch. I always assumed it was just some unrealistic simulation of motor torque effects so ignored it. Checking HITL logs though it does seem the control response is correct so it's trying to correct the roll but can't probably because airspeed is too low at the start of the launch for the controls to work effectively.

I could agree with you because everything you wrote is correct, but then why, if the aircraft configuration is the same, the autolaunch parameters are the same, and the launch method is the same since I was using INAV version 5, since version 7 everything has changed? This, to me, is proof that something has been done in the code, unfortunately making things worse.

Then, in my opinion, the INAV system has enormous potential, and I'm very sorry I haven't been able to find a solution to my other major problem, issue ID 11223 "Wobbles behavior roll axis fixed wing," which unfortunately forced me to temporarily abandon INAV until I can build the Mini Talon Pro again. I think the aforementioned issue ID 11223 deserves further investigation by the developers, because in my opinion, in some circumstances it causes the system to resonance, which is beyond the scope of parameterization, and this is a bug for me. Thank you again for your invaluable help and all INAV community.

Did you try the same wobbly wing with Ardupilot ? If so was the roll oscillation still present ? I've found flying wings to be inherently "wobbly" compared to conventional aircraft although it tends to be more an issue in pitch than roll perhaps. The oscillation on that wing looks a lot like Dutch roll which can be a problem with these wings as I understand it. Did you try physical tweaks to the model, e.g. move C.G., servo throws etc ? Also increasing cruise speed would help since wobble gets worse the slower you go.

No, Ardupilot worked well from the start because it doesn't experience resonance in the roll axis like INAV unfortunately does with this particular model, specifically the ARWING PRO. For example, with the Phantom FX-61, this behavior doesn't occur, and it's still a fixed wing, just a little larger.
I've tried everything for over a year, as explained in detail in issue ID 11223, but unfortunately this behavior can't be eliminated in the ARWIN PRO build or at least, I, who am not a beta tester, couldn't.

[sensei-hacker]

Just checking -
Was "Wobbles behavior roll axis fixed wing" also on the aircraft with the deformed control surface?

[Fabio-FPV]

Yes, but it's not relevant. If you buy an ARWING PRO and install INAV, you'll definitely experience wobbly behavior on the roll axis, especially during automatic modes like CRUISE, RTH, and COURSE HOLD. Any adjustments to the PID settings and the like won't help at all. I bought two for this very reason and tried everything for almost a year, but then, unfortunately, I gave up. It's all explained in detail in ISSUE ID 11223.

[sensei-hacker]

Hundreds of people who do NOT have a bent elevon have not reported this problem.
One person did report a problem occurred while they had a bent control surface.
Based on the evidence available at this time, I don't think we can conclusively say that planes without bent control surfaces have the same problem.

The evidence I have seen shows that WITH a bent elevon, there is a roll movement, which INAV corrects for, then it happens again and again. The only evidence I have for what happens when the elevon is properly straight is that people with good elevons haven't mentioned a problem.

If you dip the elevon into boiling water and straighten it out, THEN have a problem, that will be more interesting, to me.