Mavic Air Fly Away

[Trickydicky]

Did you happen to to have it's failsafe set mto land instead of RTH or maybe had it set to return at current altitude in which case if was 100 meters away it would land on the spot? any case sorry for your loss..

Thanks for the sentiment kidroc. The drone was set to RTH. The drone didn't land on the water.

 

[Trickydicky]

if it really did turn 120 deg and flew off at the same height (low above the watter) it would have hit the trees on the south shore.

Thanks for this. I actually thought it was high enough to get over the trees, but it's an idea, I'll go over tomorrow and have a look along the shore line.

 

[Trickydicky]

I was just wondering if it had got wet if that might have accounted for the telemetry discrepancies and disconnect. Right now there seems to be no way to reconcile your observations with the flight data, which is not a situation that I've seen before

Thanks sar104. I've been thinking that, like most electronics, if it's going to fail then it's going to fail early and maybe this is what happened here. I can't reconcile the data suggesting a crash which shows the nose pitching up and the final roll to the right, with the data suggesting the drone didn't crash i.e. it maintaining speed and descent in accordance with the appliance of down throttle.

Is there anything in the flight data that gives an indication of the direction the drone was heading at final disconnect? I guess the information must be held somewhere to plot the trace of the flight. I could use the info to try to look for the drone in the tree line at the opposite end of the lake.

Anyway I gave up getting anywhere with DJI before my first posting on here, and went out and bought a new mavic air. Despite having the thing for 10 days now I still haven't flown it. Was waiting to pluck up the courage and to see if there were any lessons to learn. I guess the main lesson is don't fly your drone over water.

 

[Trickydicky]

To me the log also looks like controlled flight into the water

What part of the log suggests that to you?

 

[Trickydicky]

It looks like you were trying to fly close to the water. You did it!

If you look at the heights as recorded you see that I flew very low over the water for a long time, then flew into the water, went a couple of feet under, came out of the water and flew further along before flying into the water again and diving to a depth of 28ft, all the time travelling at 16mph before the drone packed in. The recorded heights are wrong, I was nowhere near the surface of the water.

 

[sar104]

Thanks sar104. I've been thinking that, like most electronics, if it's going to fail then it's going to fail early and maybe this is what happened here. I can't reconcile the data suggesting a crash which shows the nose pitching up and the final roll to the right, with the data suggesting the drone didn't crash i.e. it maintaining speed and descent in accordance with the appliance of down throttle.

Is there anything in the flight data that gives an indication of the direction the drone was heading at final disconnect? I guess the information must be held somewhere to plot the trace of the flight. I could use the info to try to look for the drone in the tree line at the opposite end of the lake.

Anyway I gave up getting anywhere with DJI before my first posting on here, and went out and bought a new mavic air. Despite having the thing for 10 days now I still haven't flown it. Was waiting to pluck up the courage and to see if there were any lessons to learn. I guess the main lesson is don't fly your drone over water.

The final yaw value in the log, after the reconnect, was 157°, compared to 97° just before. Is that consistent with what you saw? Given the reported roll at that point is 67° I'm not sure that we can put much faith in any of those numbers however, if the aircraft really were still airborne.

 

[Jeff7577]

If you look at the heights as recorded you see that I flew very low over the water for a long time, then flew into the water, went a couple of feet under, came out of the water and flew further along before flying into the water again and diving to a depth of 28ft, all the time travelling at 16mph before the drone packed in. The recorded heights are wrong, I was nowhere near the surface of the water.

Sorry, you have provided 0 evidence to support your case. Where is the cached video of this flight? IMU altitude is not perfect, but it’s good enough to show you were skimming the water on purpose. No surprise that’s where it ended up.

I have never seen the data be as skewed as you think it is, and that’s because it isn’t. You put it in the water.

 

[laurens23]

What part of the log suggests that to you?

The log you were flying in the direction of the Homepoint, just as you described.
roughly 10 seconds before the connection is lost you are flying full forward elevator (green) and do some full up/down throttle (blue).
moments before the disconnect you are low above the water (red), (the sensor might be off, but it was a few meter max) and you go from full forward and down to neutral and still full down (@153 sec).
during a braking manoeuvre the drone can lose some altitude. together with the downward momentum it still had, and the fact it disconnected and not returned home, it seems likely it just hit the water.
after that there are a few readings that are highly improbably if the drone were still flying (the huge roll angle @sar104 pointed out)

 

[Trickydicky]

I have never seen the data be as skewed as you think it is, and that’s because it isn’t. You put it in the water.

Thanks Jeff7577. So sar104 thinks the drone hit the water at 353.7 but the drone keeps moving forward at the same speed, continues to descend at the same rate and moves a distance of 37ft after first impact. Can you explain this, or do you have a different view of when the drone hit the water?

 

[sar104]

Thanks Jeff7577. So sar104 thinks the drone hit the water at 353.7 but the drone keeps moving forward at the same speed, continues to descend at the same rate and moves a distance of 37ft after first impact. Can you explain this, or do you have a different view of when the drone hit the water?

Just to be clear - I don't "think" that happened - I simply read and interpreted the log data. And then I tried pretty hard to reconcile your observations with those data. But if you would prefer to figure this out for yourself without any input then that's just fine too.

 

[Trickydicky]

moments before the disconnect you are low above the water (red), (the sensor might be off, but it was a few meter max) and you go from full forward and down to neutral and still full down (@153 sec).

Thanks laurens23, you've provided me with even more information which doesn't make sense. I hadn't realised that I had started to slow the drone down just before the disconnect. The flight log shows that the drone is moving forward at about 16mph at 353. A second later when the forward elevator is at zero the drone is still flying at 16mph, so it hasn't slowed and doesn't start slowing until 354.2. At 354.7 it is still doing 12.1mph and at 5.56.3 it is still flying at 8.5mph. Is it normal to get a lag in the drone responding, and why was the drone still flying forward at 8.5mph when there had been no forward elevator applied for 3.3 seconds. I accept that the possibility of the drone still being in the air if the roll angle was 67 degrees is remote. But it is only one reading. All the other readings support the drone still being in the air. So maybe the roll angle reading was spurious because the drone had developed a fault. Even the nose pitching up, which sar104 though was when the drone hit the water could be explained if the drone was slowing, the nose would naturally pitch up if application of forward elevator was stopped (particularly if there was a headwind). So the only reading suggesting the drone was not flying is the roll angle.

My problem is, and despite what you all keep saying, the drone didn't hit the water, it disconnected, turned and flew off.

 

[Trickydicky]

Just to be clear - I don't "think" that happened - I simply read and interpreted the log data. And then I tried pretty hard to reconcile your observations with those data. But if you would prefer to figure this out for yourself without any input then that's just fine too.

Sorry sar104, I was just quoting your first analysis, of course I'm looking for help, that's why I posted the flight log, I clearly didn't do it to make friends! (poor Scottish humour)

 

[Trickydicky]

Where is the cached video of this flight?

Sorry Jeff7577 I should have picked this up because it is something that really annoyed me when dealing with DJI. I spent best part of a month talking to these people and they never asked that question. I only found out about the possibility of a cached video looking through this forum. Of course there isn't a video because there is no space on my phone.

 

[Trickydicky]

moments before the disconnect you are low above the water (red), (the sensor might be off, but it was a few meter max) and you go from full forward and down to neutral and still full down (@153 sec).

Sorry laurens23 but I've been looking at the flight data again and it shows that when I start reducing the forward elevator at 353 the drone is flying at 15.1mph and as the forward elevator reduces the drone actually speeds up to 16.1mph at 353.6. How can this happen, there is something seriously wrong with the data.

I have to ask how do you get access to all the extra data, when I go into phantom help I can only see a very limited data set.

 

[Trickydicky]

The final yaw value in the log, after the reconnect, was 157°, compared to 97° just before.

Thanks for this sar104. I was at the park today trying to figure out where the drone went. Being on site again I'm pretty certain the drone turned nearer 160 degrees before heading off. As laurens23 said the drone wasn't high enough to clear the trees at the far end of the lake. So if it didn't automatically avoid them then the drone is now in the water because all the trees overhang the bank and the drone definitely isn't hanging in any of the trees.

 

[MavicCF]

I would say that you are dealing with two very smart and capable people (@laurens23 and @sar104). They both know what they are doing and sar104 has handled many hundreds of these cases without any mistakes (that I have heard of). The log data has been proven correct time and time again, and even if there was a problem with it, these guys would find it.
Honestly if they say it went into the water when it did, than the likelihood of that being the case is extremely high. These guys are experts, and the logs just don't lie! -CF

 

[sar104]

Thanks for this sar104. I was at the park today trying to figure out where the drone went. Being on site again I'm pretty certain the drone turned nearer 160 degrees before heading off. As laurens23 said the drone wasn't high enough to clear the trees at the far end of the lake. So if it didn't automatically avoid them then the drone is now in the water because all the trees overhang the bank and the drone definitely isn't hanging in any of the trees.

Unless you find the aircraft I'm afraid that this is going to remain a mystery. The pertinent flight data are mostly reasonable, with a couple of unexplained features:



Pitch follows the elevator input up to the first disconnect. Altitude follows your throttle input. Roll doesn't vary much, as expected without much wind and no aileron input. Velocity east (y-speed) behaves as expected when the elevator is centered. Velocity north (x-speed) does show an anomalous increase (to the south) at that time but that's likely due to the slight roll right and is not uncommon on sudden deceleration.

So until that disconnect I see nothing particularly unexpected in the data. And given that the altitude reading suggests a splash down the following single data points are inconclusive. Without your description of what happened it looks like a water impact.

 

[sar104]

I would say that you are dealing with two very smart and capable people (@laurens23 and @sar104). They both know what they are doing and sar104 has handled many hundreds of these cases without any mistakes (that I have heard of). The log data has been proven correct time and time again, and even if there was a problem with it, these guys would find it.
Honestly if they say it went into the water when it did, than the likelihood of that being the case is extremely high. These guys are experts, and the logs just don't lie! -CF

The complication is that the OP is quite certain that he saw it fly away, and it's hard to discount such direct, simple eye-witness testimony. So either he was mistaken, having consumed too much lowland whisky perhaps, or something very strange happened that the log file doesn't explain.

 

[Drone062074]

My Mavic Air was showing disconnected and I was able to still fly it. So I’m sure you thought maybe it wasn’t doin anything when it showed disconnected but it was actually still operating ok. I switched to using the bottom of controller and the disconnected warnings haven’t came up.

 

[albatross911]

Could it simply be the wrong log?