Mavic went insane

[Bzekovic]

I just had an accident using a Mavic in Litchi vr mode - my drone went completely unsane: it started spining uncontrolably and then started ascending really fast until it hit some tree branches. How can I found out who's fault is it...and does it evem matter?
Drone itself doesn't look damaged from thr outside, but gimbal mottor doesn't seem to work correctly and I get some errors. Any advice what to do? Thanks in advance

 

[sar104]

Check the DAT file, or post it here for further analysis.

How to retrieve a V3 .DAT File

 

[Bzekovic]

Unfortunately, it turned out that I forgot to put the sd card back into the drone before the flight, so I guess none that was recorded (?) Now I'll never know what happen. It is really strange that nobody experienced similar problems...two times I tried to use Litchi vr - first time it completely blocked my remote (I had to restart it to gain controle of drone again) and now this. Thanks for your reply anyway.

 

[msinger]

The DAT flight log is stored on the Mavic's internal memory card. That'll occur whether or not the removable memory card is in your Mavic.

 

[Bzekovic]

Ok, here is a DAT file. I don't know if it matters, but I noticed a bit unusual behaviour in vr mode - when I send it forward it keeps going forward, even if I release a stick. I need to give him back command to make it stop. This time it started turning in a similar way, like it received a command and waits for a stop, but doesn't get it and then it lounched itself into the branches. Drone wasn't reacting to my remote commands at that time. If I have to guess what happen I would say that Litchy messed up my remote again and then drone went into RTH mode.

FLY034.DAT

 

[BudWalker]

Ok, here is a DAT file. I don't know if it matters, but I noticed a bit unusual behaviour in vr mode - when I send it forward it keeps going forward, even if I release a stick. I need to give him back command to make it stop. This time it started turning in a similar way, like it received a command and waits for a stop, but doesn't get it and then it lounched itself into the branches. Drone wasn't reacting to my remote commands at that time. If I have to guess what happen I would say that Litchy messed up my remote again and then drone went into RTH mode.

FLY034.DAT

This isn't the right .DAT. There was no motor start and the satellite count was low. Maybe done inside? The .DAT was created at time 2017-10-7 18:22:54 local. Take a look here for help on picking the right .DAT

 

[Bzekovic]

Actually, that was the oldest file from today that I could find. I checked all the files and it seems that there is no record from that time and that Mavic uses SD card for logging after all.
I managed to sync logs from Litchi to AirData, but I doubt that it could be useful.

 

[sar104]

Actually, that was the oldest file from today that I could find. I checked all the files and it seems that there is no record from that time and that Mavic uses SD card for logging after all.
I managed to sync logs from Litchi to AirData, but I doubt that it could be useful.

The Mavic doesn't use the removeable SD card for flight logs - they are on the internal card. And that's easy to verify, since you will find no DAT files on the removeable card and you will find them on the aircraft when no removeable card is present.

 

[Bzekovic]

Ok, one more try. I was looking at the time stamps on DJI Assistant and none of the logs with that date have motor activity and also times don't match with a flight. This log is the last one that I could find that has an actual flight in it, but it has a date of my previous flight, which could only be true if the file was was created automatically when the last one was finished. All the flights that I downloaded are in sequential order so I guess there is nothing missing (I was afraid that maybe the file got corrupted because of the file).
I can also export Litchi Log from AirData if that helps. According to it, crash time is around 16:59 on 7th of October.
There are also some blackobox files, but I don't know if they would be useful to anybody.

FLY029.DAT

 

[sar104]

So that looks like the correct log - it behaves somewhat as you described, and then climbs and hits something at 442 seconds. It is a very strange flight profile:



So you were flying around in some kind of VR mode? This is very close to the ground, and includes a couple of periods of severe yaw excursions. I cannot see the causes of any of those maneuvers - it wasn't direct stick inputs. Can you elaborate on how you were controlling the aircraft and what you were trying to do?

 

[BudWalker]

Yeah, this is a strange one. I was going to guess that the Mavic got caught in an updraft. The altitude gain occurred before the erratic flight. It was flying along very close to the ground and then started rising with no throttle at first. It then continued to rise with full negative throttle.


At time 291.624 the AC declared a ctrl failure, from the eventLog stream.
291.624 : 1996251525 : 21920 [L-RC]craft ctrl failed!!!
This was after it rolled left uncommanded


The only thing I can think of is that it was caught in an updraft that then got stronger on the right side.

Following this is when the motors started trying to compensate causing the erratic flight.

 

[Bzekovic]

Actually, I was just hovering in my backyard, trying Litchi vr mode, when some "noise" started - drone was making some moves that had nothing to do with my head movements, so I took the goggles of to see what is going on and it was still rotating. If I remember correctly, it began turning even quicker and than it started ascending really fast and hit the branches, even though I instructed it to go down. I think it rolled over and hit the ground full speed with it's top side. In my previous try with Litchi vr, remote control got completely stuck at one point, but I managed to regain controle by restarting it. I can only imagine that something similar happen and that drone automatically tried to return home (by ascending first).

 

[BudWalker]

Actually, I was just hovering in my backyard, trying Litchi vr mode, when some "noise" started - drone was making some moves that had nothing to do with my head movements, so I took the goggles of to see what is going on and it was still rotating. If I remember correctly, it began turning even quicker and than it started ascending really fast and hit the branches, even though I instructed it to go down. I think it rolled over and hit the ground full speed with it's top side. In my previous try with Litchi vr, remote control got completely stuck at one point, but I managed to regain controle by restarting it. I can only imagine that something similar happen and that drone automatically tried to return home (by ascending first).

I don't know much about VR mode. Isn't just the gimbal that can be controlled with head movements? In particular, throttle isn't controlled in VR mode? The reason I'm asking is that altitude gain doesn't seem to be the result of increasing either motor commanded or speed.

Is it possible you weren't in VR mode but some other mode where the AC itself is controlled? When the switch to VR mode happened there was an immediate change in Yaw commanded by the motors but no rudder input

 

[Bzekovic]

In Litchi mode you can control gimbal and, if you select Imersive mode, it can also folow your head left and right turns. No throttle controle. I suspect that the throttle and elevation gain happen from drone going to RTH mode because it sensed some failure. Similar behaviour happen to me when the battery was low, but I managed to battle it, because rc was functional - drone was just hovering and than, all of a sudden, started ascending full throttle. I reread the manual and found out that the standard RTH procedure is to ascend some 20-30m and then to start return to the home point. That is the only reason that made sense to me.

 

[BudWalker]

In Litchi mode you can control gimbal and, if you select Imersive mode, it can also folow your head left and right turns. No throttle controle. I suspect that the throttle and elevation gain happen from drone going to RTH mode because it sensed some failure. Similar behaviour happen to me when the battery was low, but I managed to battle it, because rc was functional - drone was just hovering and than, all of a sudden, started ascending full throttle. I reread the manual and found out that the standard RTH procedure is to ascend some 20-30m and then to start return to the home point. That is the only reason that made sense to me.

RTH wasn't initiated. And, if it was it would have just descended since it was within 20 meters of the HomePoint.

 

[Bzekovic]

I found that very unusual also, but it was the only thing that could explain this behaviour. First time (battery low), it was closer than 20m as well (almost the same spot as this time) and didn't have any other issues. It looks like this ascend is done automatically, but, if it's not RTH, I don't have a clue why. What do you think I should do about it?

 

[BudWalker]

I found that very unusual also, but it was the only thing that could explain this behaviour. First time (battery low), it was closer than 20m as well (almost the same spot as this time) and didn't have any other issues. It looks like this ascend is done automatically, but, if it's not RTH, I don't have a clue why. What do you think I should do about it?

We could take a look at that flight as well f you can retrieve the .DAT.

I don't have any advice since the cause of the incident hasn't been determined

 

[Bzekovic]

I doubt I could find that, it was weeks ago. I'll try contacting DJI and send them these logs. Thanks for your help anyway

 

[Bzekovic]

Quick update (maybe someone would find it useful) - I contacted DJI support and after a long correspondace that really pissed me off, I managed to get this graph from them. It appears that dron received an UP command from SDK (Litchi) and I have no idea how can that happen in VR mode. No response from Litchi yet.

 

[Oingo Boingo]

U had the drone set to immersive mode hence the signal came from SDK in Lichi. If u tilted your head up it would literally fly up at a rate of knots. If you til your head left it would move left and so on. It would be the only explanation. I've been flying Lichi since I've had my drone and never had any issues. Used the VR mode as well but turned immersive mode off as I didn't want to get into head tracking.... I prefer flying with the sticks