https://www.safecom.gov/safecom/26-0936
I am fairly curious if their interpretation of the crash being cause by a updraft and the aircraft disarming is truly the root cause. Unfortunately this aircraft was not recovered. I have had some interesting stuff happen with DJI back in the day when doing some studies on RX fire for my masters degree but never to this extent and we had plenty of convective air. To my knowledge this is the first safecom that has been issued for something like this.
Not that it really makes any difference, but the Alta X runs PX4 not ArduPilot.
Well aware of that thank you, my intent was to start a discussion. They are calling this a bug, could ardu exhibit something similar? I am just curious of others ideas. We have operated in some very harsh environments and aside from some bucking and jumping around in the convective column have never experienced anything like this using ardu or any other flight controller software. I also can’t imagine that there was enough convective force on this incident to cause a alta x to (float) and think it had landed and needed to disarm
I’m not overly familiar with PX4, but a little code crawling reveals that auto disarm happens in roughly the same types of scenarios as ArduPilot. It seems quite incredulous that an updraft would be stable enough to satisfy the landing detection algorithm, and I think it’s more likely that there was an upset (excess attitude failsafe) or other system failure that may have resulted in a flight termination failsafe. It would be very telling to see a GCS/telemetry log or copy of the messages emitted prior to disarm to see if “Landing detected” or some other failsafe occurred.
If an updraft could cause a near perfect altitude hold with min throttle for several seconds, ArduPilot would suffer the same fate, but that seems very unlikely.
You know the ardu code much better and I was just starting my initial poking around into both to see what similarities and differences they had. I appreciate that! I cannot imagine this aircraft being in a steady enough state to make it think it was landed. The thing to me that is probably telling is they mention the camera getting vibrations prior to the all happening, makes me think too like you mentioned @Yuri_Rage there may have been something else in the background going on that is the initial cause. The aircraft unfortunately burned up so any post analysis is pretty hard to do.
Most GCS software logs telemetry. That would potentially be available.
Per the safecom, “The Pilot in Command and Visual Observer immediately reviewed the flight log on the controller display and observed that the controller had displayed “Takeoff Detected” continuously throughout the flight. This indication was not recognized by either crewmember during flight operations and was only identified during the immediate review following the mishap.”
“After manufacturer inspected the flight log it was determined a bug in the software prevented the audible updraft warning from triggering. A upcoming firmware release should correct the false landing detection.”
Considering both auto disarm detection are similar between ardu and px4, I would be curious what they determined what this bug was. It looks like they were able to recover some of the aircraft and perform some analysis.
It would be an edge case in the PX4 code if there is actually a bug. There isn’t really anything shared between PX4 and ArduPilot in that regard.
The entire report is a bit suspect given the observation of camera shake prior to the event, as you read.