Lost control, then 2 seconds later lost connection, on my Mavic Pro what happened ??

[BudWalker]

@BudWalker - any thoughts on this event?

I suspect it's because the FC didn't think this was a "normal" landing and the sequence is different. The FC thought it was already on the ground when the throttle went full negative.

I looked at the landing sequence in an MP .DAT. After full negative throttle and no descent the motors are more or less constant speed for 2 seconds. Then 1 sec of reduced speed followed by full stop. But, the FC only thought it was near the ground as opposed to on the ground.
166.461 : 17692 [L-FMU/FSM]near ground
167.385 : 17738 [L-FMU/MOTOR]safe_near_grd:true

In this event the FC thought it was on the ground

95.291 : 8465 [L-FMU/FSM]state changed. cur: on ground
95.320 : 8467 [L-FMU/MOTOR] Stop. reason:landing.RC_Thr

 

[Goin Up]

This looks like it should have been a pre-programmed Litchi flight.

 

[mikeyway]

This looks like it should have been a pre-programmed Litchi flight.

No I don’t really trust third party applications. If you look at the input data it was all manual.

 

[mikeyway]

I suspect it's because the FC didn't think this was a "normal" landing and the sequence is different. The FC thought it was already on the ground when the throttle went full negative.

I looked at the landing sequence in an MP .DAT. After full negative throttle and no descent the motors are more or less constant speed for 2 seconds. Then 1 sec of reduced speed followed by full stop. But, the FC only thought it was near the ground as opposed to on the ground.
166.461 : 17692 [L-FMU/FSM]near ground
167.385 : 17738 [L-FMU/MOTOR]safe_near_grd:true

In this event the FC thought it was on the ground

95.291 : 8465 [L-FMU/FSM]state changed. cur: on ground
95.320 : 8467 [L-FMU/MOTOR] Stop. reason:landing.RC_Thr

I sadly do not have a .dat viewer, but I know it also records VPS data. And when landing, the MP usually takes it into account and I assume in the state „near_ground“ it sends a request for clearance to the VPS. If it return positive it changes the Boolean „safe_near_ground” to true.
This is probably only during a automatic landing and not during a manual of hand catch. I still think it’s a big design flaw not to take those sensors into account, maybe at least give a warning if the FC will shut down the motors without any correct VPS information present.
I’ll pass those findings onto DJI, let’s hope they have a open ear for hobby pilots.

Ps. If the mods could maybe change the title of the thread to : “drone shuts off motors in the mountain due to false interpretation of thermal winds” (or something similar) that people with the same experience will easier find this thread on google etc.

 

[sar104]

I sadly do not have a .dat viewer, but I know it also records VPS data. And when landing, the MP usually takes it into account and I assume in the state „near_ground“ it sends a request for clearance to the VPS. If it return positive it changes the Boolean „safe_near_ground” to true.
This is probably only during a automatic landing and not during a manual of hand catch. I still think it’s a big design flaw not to take those sensors into account, maybe at least give a warning if the FC will shut down the motors without any correct VPS information present.
I’ll pass those findings onto DJI, let’s hope they have a open ear for hobby pilots.

Ps. If the mods could maybe change the title of the thread to : “drone shuts off motors in the mountain due to false interpretation of thermal winds” (or something similar) that people with the same experience will easier find this thread on google etc.

The VPS system did not see the ground in this case.

 

[mikeyway]

The VPS system did not see the ground in this case.

What what was the variable „safe_near_ground_“ set to ?

 

[sar104]

What what was the variable „safe_near_ground_“ set to ?

It had not been changed since just after takeoff:

3.361 : 3869 [L-FMU/MOTOR]safe_near_grd:false

 

[BudWalker]

What what was the variable „safe_near_ground_“ set to ?

It would appear to be set to false. The last time it was set was right after launch at 3.361 secs
3.361 : 3869 [L-FMU/MOTOR]safe_near_grd:false

 

[Goin Up]

No I don’t really trust third party applications. If you look at the input data it was all manual.

Manual is why you lost contact with your drone. You went over the hill and lost transmission link. If you were using litchi the programming would have been uploaded to your drone and your controller wouldn't have any effect. The Drone would have completed its Mission back at your location.

 

[sar104]

Manual is why you lost contact with your drone. You went over the hill and lost transmission link. If you were using litchi the programming would have been uploaded to your drone and your controller wouldn't have any effect. The Drone would have completed its Mission back at your location.

No that's not what happened. The FC shut down the motors and the aircraft fell to the ground. It lost connection on the way down - that was unrelated to the cause of the crash.

 

[mikeyway]

Manual is why you lost contact with your drone. You went over the hill and lost transmission link. If you were using litchi the programming would have been uploaded to your drone and your controller wouldn't have any effect. The Drone would have completed its Mission back at your location.

I'm with @sar104 on this one. The drone's state would have been set to "on ground" anyway since there was no downward motion while pulling back on the throttle. I'm not sure how much a "mission" has control over the FC but I doubt that it's able to force to keep the motors running.

 

[ylmavicpro]

Here's a more accurate depiction of the flight, using just GPS location and height.

View attachment 70084

sorry for my technical ignorance: what does the yellow, rep. red curve mean? I try to understand and profit more from reading and learning in the forum. I admiring your knowledge and kindness to help here always. Thanks.

 

[mikeyway]

sorry for my technical ignorance: what does the yellow, rep. red curve mean? I try to understand and profit more from reading and learning in the forum. I admiring your knowledge and kindness to help here always. Thanks.

No worries, everyone starts small and I’m also just getting into flight analysis. The red line seems to be the movement over ground, the yellow line the movement in the air.
I’m hoping sar gives out a in depth tutorial.

 

[sar104]

sorry for my technical ignorance: what does the yellow, rep. red curve mean? I try to understand and profit more from reading and learning in the forum. I admiring your knowledge and kindness to help here always. Thanks.

I should have explained that better. In post #4 the red track is the recorded flight track (IMU position + IMU height) from the txt log. The obvious problem with the data is that when the FC cut the motors the barometric height was reset to zero (that always happens) and so there is a discontinuity in the height record. The green trace is just the ground track to help visualize the path.

To fix the height problem I extracted the GPS position and GPS altitude MSL from the DAT file, and those are shown in post #18 as the more accurate yellow flight track that doesn't suffer from the height discontinuity when the motors stop. Confusingly I changed the color of the ground track to red, but it's basically the same as the green ground track in post #4.

 

[ylmavicpro]

thanks for the further explanation: sar104 and mikeyway!

 

[Bohica]

I have not looked at any of the files, but I have a few wierd experiences with my mavic 2 pro and I discovered something the other day while flying in Vietnam.

the day before my flight, I flow within the town of Dalat with no problems, everything went perfect. The next day I went to my family's coffee farm to film there new strawberries. I had alot of problems connecting to the fly cam. It would connect, then disconnect. I finally thought I connected, but on takeoff it started to fly in its own, I was able to land safely. I then decided to put my phone in airplane mode, no problems, strong connection, and finished a great flight.

I think if you have weak cell signal, or in a rural area, you need to go to airplane mode. I may always download maps and fly in airplane mode on my phone in the future from this experience.

 

[sar104]

I have not looked at any of the files, but I have a few wierd experiences with my mavic 2 pro and I discovered something the other day while flying in Vietnam.

the day before my flight, I flow within the town of Dalat with no problems, everything went perfect. The next day I went to my family's coffee farm to film there new strawberries. I had alot of problems connecting to the fly cam. It would connect, then disconnect. I finally thought I connected, but on takeoff it started to fly in its own, I was able to land safely. I then decided to put my phone in airplane mode, no problems, strong connection, and finished a great flight.

I think if you have weak cell signal, or in a rural area, you need to go to airplane mode. I may always download maps and fly in airplane mode on my phone in the future from this experience.

It must have been the strawberries.

 

[Deleted member 94047]

This is by far the weirdest crash incident I have read here. Flight computer decides aircraft has landed and cuts off motors while it is 120 meters in the air? WTH! OP, you have my deepest sympathies, I hope you take it up with DJI and that they will address it in future product development at the very least (it would be noble of them to give you a replacement too).

Out of curiosity, and for caution, how are the rotor motors cutoff when manually landing - is it the amount of seconds you have pushed down on the left stick combined with the absence of descent for a set time that triggers the motor cut off? Does the downward infrared sensor come into the equation at all? Or is it too complex and too contingent on several factors to describe here? I would greatly appreciate any explanation of how this works - and particularly the parameters that determine rotor cutoff.

Edit: Forgot to say thanks

 

[mikeyway]

Hey Doppler
Thx for your empathy, the drone truly made some awesome videos and it was a real conversation starter. I have contacted dji on Thursday, no reply yet. I think it should be the least to give me a replacement or in store credit at the total cost of the drone when I bought it. They are actually lucky that it was in the mountains and nobody got hurt or any damage was done, since in Germany they would have been liable.

As far as the crash the main problem is that it’s too easy to describe. The fc thought it was on the ground because the drone didn’t loose altitude when it should have.
There where no sensors consulted since they would have all reported that the drone was not near the ground. There is an actual value in the system that can either be true or false if the drone is near the ground. But it seems like the flight controller didn’t take that into account since it was set to false.

I’m actually a bit disappointed with Dji here. The drone is usually amazingly stable and even flies when half a rotor is missing. But not taking your sensors into account is just lazy.

The only thing I COULD imagine is that they did it on purpose to not consult the bottom facing sensors. When you land manually it’s often a hand catch and it would take longer for the drone to decide if it was actually on the ground.

 

[Deleted member 94047]

As far as the crash the main problem is that it’s too easy to describe. The fc thought it was on the ground because the drone didn’t loose altitude when it should have.
There where no sensors consulted since they would have all reported that the drone was not near the ground.

That right there is what I find most bizarre. I would have thought, at the very least, the down facing infrared sensor would be involved in determining whether the aircraft has landed or not. Making this determination solely based on the presence or absence of altitude loss is quite simply unbelievable.

I presume this was a Mavic Pro; can this possibly happen to other models in the Mavic line as well?

Edit: fixed quotation