Get more from your Mavic
Friendly, Helpful, Knowledgeable Mavic Community
Join Us Now

Mavic 2 Pro Active Track flyaway, totaled!

Zoogy

New Member
Joined
Sep 22, 2020
Messages
4
Reactions
0
Age
53
Location
New Orleans, LA
A few weeks ago I was flying my Mavic 2 Pro with a friend. I flew it once or twice in Active Track with my friend walking away from me a short distance, maybe 20 feet, just to show him how it works. Then I exited active track and flew the drone back to my position.

After a few very short flights back and forth between me and my friend (in P-GPS mode), the drone was hovering before me for a few seconds, and then it suddenly took off diagonally to the left and crossed the street, which was full of moving cars. The Mavic barely escaped hitting a moving car and then crashed into a tree. I tried to gain control, pushing up on the stick for altitude and hard left to get it out of the street, but my stick movements did not seem to have much affect on the drone.

I am certain that I did NOT put the drone into Active Track mode before the flyaway.

The gimbal/camera and forward facing sensors need to be replaced.

I purchased the drone just under a year ago, so it is still under warranty and I was hoping that DJI would analyze the logs and offer to cover repairs. They just responded that the drone was operating normally in Active Track mode and therefore they would not cover repair costs.

I am posting the log here, hoping to get some feedback from the forum on what happened. HOW did it go into Active Track mode? And WHAT was it tracking? There is no way that a person, animal or object crossed the street at that time. Why did it suddenly take off, away from me and my friend, and how do I convince DJI that it was NOT "working as designed"?

Here is my log at Phantom Help: DJI Flight Log Viewer - PhantomHelp.com

Thanks for taking a look!

PS - here is the response from DJI tech support:

Dear Daniel,

.
.
.

FLY050

1. Unit was in ACTIVE TRACK mode and was working as designed.
2. At t=461s, relative height=1.7m unit recorded external impact.
3. User commanded ROLL left prior to impact.
4. Map matches crash location with trees. Tree branches, Cables, Light poles, slanted roof, poor lighting. This is not detected by vision system due to hardware limitation.

Force impact date: 2020-08-19
Force impact GPS location: 29.9303359, -90.1170812

According to the analysis, the incident was not caused by any product malfunction.

We can still proceed with the repair, however, you will cover the repair charges.
 

Meta4

Well-Known Member
Joined
Oct 30, 2016
Messages
11,368
Solutions
1
Reactions
13,363
Age
66
I am posting the log here, hoping to get some feedback from the forum on what happened. HOW did it go into Active Track mode? And WHAT was it tracking? There is no way that a person, animal or object crossed the street at that time. Why did it suddenly take off, away from me and my friend, and how do I convince DJI that it was NOT "working as designed"?
About all anyone can tell from looking at the flight data is that from 3:52.9 you centred the joysticks and left the drone hovering a little above head height
At 4:02.9 the flight mode was changed from P-GPS to Active Track and at 4:05.6 it started to fly backwards to the south (across the road), and then west, parallel to the road, accelerating to 17.7 mph (at 4:17.1) and then beginning to slow down.
At 4:13.4 you started to give joystick input with throttle, rudder and aileron to vary the flight path.
At 4:16.7 you pushed the right stick hard to the left til 4:18.3.
At 4:18.2 there is a warning of an upward obstacle detected.
At 4:18.3 the pitch, roll and yaw data indicates a collision.

It's not clear what the drone was tracking or if it would have avoided the crash on its own.
Your full stick move might have contributed to the collision.

For future reference, the quickest way to cancel any automated flight (like Active Track, RTH etc) is to flick the flight mode switch to Sport and back to P-GPS.

 

Zoogy

New Member
Joined
Sep 22, 2020
Messages
4
Reactions
0
Age
53
Location
New Orleans, LA
About all anyone can tell from looking at the flight data is that from 3:52.9 you centred the joysticks and left the drone hovering a little above head height
At 4:02.9 the flight mode was changed from P-GPS to Active Track and at 4:05.6 it started to fly backwards to the south (across the road), and then west, parallel to the road, accelerating to 17.7 mph (at 4:17.1) and then beginning to slow down.
At 4:13.4 you started to give joystick input with throttle, rudder and aileron to vary the flight path.
At 4:16.7 you pushed the right stick hard to the left til 4:18.3.
At 4:18.2 there is a warning of an upward obstacle detected.
At 4:18.3 the pitch, roll and yaw data indicates a collision.

It's not clear what the drone was tracking or if it would have avoided the crash on its own.
Your full stick move might have contributed to the collision.

For future reference, the quickest way to cancel any automated flight (like Active Track, RTH etc) is to flick the flight mode switch to Sport and back to P-GPS.

Thanks, Meta, for taking a look. Also thanks for the tip on quickly cancelling automated flight, I will remember that! At the time of this mishap, I was not even aware that the drone was in Active Track mode.

Why would the drone have accelerated to 17.7 mph when the sticks were both in the zero position? Why did the drone even fly off in the first place, and how could it be following something in Active Track mode if it was flying backwards? How can you draw a square around a target that is behind the drone and therefore not visible on the screen? I for sure did not hit GO or START or any other command on the screen to tell it to begin flying.

Drone flew across the street. If it was following a target, that target would have had to cross the street (full of moving cars), or be across the street already (in shade and heavy foliage) and if flying backwards how could it even see such a target?

When you combine the flight data with the map, is it not apparent that the thing took off on its own with no input from me and no target to follow?
 

Meta4

Well-Known Member
Joined
Oct 30, 2016
Messages
11,368
Solutions
1
Reactions
13,363
Age
66
Sorry but once it's in an automated flight mode, we can't tell much about what automated controls the drone is reacting to, so what answers I can give are limited..
Why would the drone have accelerated to 17.7 mph when the sticks were both in the zero position?
Because it was trying to track something (I can't tell what).
In active track, joystick input isn't necessary.
The drone takes control input from the app and its optical tracking.
See the lack of joystick input when you flew in Active Track earlier from 2:40.2 or 2:48.3 for examples.
Why did the drone even fly off in the first place, and how could it be following something in Active Track mode if it was flying backwards? How can you draw a square around a target that is behind the drone and therefore not visible on the screen? I for sure did not hit GO or START or any other command on the screen to tell it to begin flying.
When you combine the flight data with the map, is it not apparent that the thing took off on its own with no input from me and no target to follow?
The data doesn't give enough information to tell.
But it does show the flight mode changed to Active Track from 4:02.9 after a period of hovering.
 
  • Like
Reactions: _gonzales_

slup

Well-Known Member
Premium Pilot
Joined
Oct 18, 2019
Messages
3,768
Reactions
8,926
Age
54
Location
Sweden
As already said, when automated flight modes is used it's hard to pinpoint all needed to say if it is a pilot or a AC system error. In this case I think it can be useful to also look into the mobile device .DAT log, it's stored in your mobile device in a subfolder, MCDatFlightRecords, in the folder that contained the TXT log you uploaded to Phantomhelp.com. The correct .DAT log ends with FLY050.DAT, try to retrieve it & attach it here.

It's mainly 2 thing's that perhaps can be confirmed ... firstly, we see that the AC yaw 20 degrees immediately after the last active track mode is initiated, that can indicate that something starts to get followed. If we are lucky something in the DAT event log can show something about that.

The blue graph show the yaw angle ... and the blue background the active track mode (the black - heading speed needs to be taken times 10 to be correct).
1600846545570.png

Secondly the take off spot & the TXT event log make me suspicious ... can we be dealing with a yaw error due to a magnetic disturbed take off spot kicking in late in the flight due to the low flight path with little need for the FC to hold/correct position?

This is the flight area & the take off spot ... a lot of magnetic thing's every where on ground level with those tram tracks.
1600846995398.png
1600847013092.png

Then we have this in the TXT event log ...
1600847069213.png

In retrospect ... this area wasn't the best place to fly a drone, especially if it was a lot of traffic there.
 

RichR

Well-Known Member
Joined
Apr 7, 2020
Messages
79
Reactions
89
Location
North Yorkshire, UK
Also, it may well have been illegal to fly there anyway - most countries have a 'not within a certain distance of people, vehicles or buildings not in your control' regulation. There's a lot of things you could hit that have a high possibility of damaging your drone or something resulting in you being taken to court.
 
  • Like
Reactions: kidroc

Zoogy

New Member
Joined
Sep 22, 2020
Messages
4
Reactions
0
Age
53
Location
New Orleans, LA
What stands out most to me is the compass interference. I will agree that there could have been a lot of electro-magnetic noise nearby. However, I had flown back and forth a few times in this spot without any issue, and nothing had changed - there were not any street cars that passed by during that time, and I stayed in roughly the same position the entire time. It must have lost its position, maybe even thought it was avoiding non-existent obstacles?

I still don't understand how the drone switched to Active Track mode. I definitely did not switch to Active Track by navigating to it in the DJI Go app on my iPad. If that had been the case, I would have been scratching my head trying to figure out what it was following, and my complaint to DJI would have been about Active Track not working. It was a surprise to see it in the logs. Somehow after I used it for one short flight and then switched back to P-GPS, it switched itself back to Active Track.

Anyways -- the quote just came in from DJI and it is not nearly as bad as I was expecting. They do not have to replace the camera, just the gimbal axis arm assembly and the forward vision unit, total is under $400. I'm just going to pay it and get it repaired.

Thanks everyone for your input!
 

slup

Well-Known Member
Premium Pilot
Joined
Oct 18, 2019
Messages
3,768
Reactions
8,926
Age
54
Location
Sweden
Electromagnetic noise will not cause your AC to stear off & crash ... at most it will lag your AC-RC connection, it's what near your AC when it powers up that matters. Really think a look into the .DAT log from your mobile device is worth the effort, maybe we find something that can serve as a good learning for you & all the members here at the forum.
 

Deleted Member 997

Well-Known Member
Joined
Sep 17, 2020
Messages
156
Reactions
103
As already said, when automated flight modes is used it's hard to pinpoint all needed to say if it is a pilot or a AC system error. In this case I think it can be useful to also look into the mobile device .DAT log, it's stored in your mobile device in a subfolder, MCDatFlightRecords, in the folder that contained the TXT log you uploaded to Phantomhelp.com. The correct .DAT log ends with FLY050.DAT, try to retrieve it & attach it here.

It's mainly 2 thing's that perhaps can be confirmed ... firstly, we see that the AC yaw 20 degrees immediately after the last active track mode is initiated, that can indicate that something starts to get followed. If we are lucky something in the DAT event log can show something about that.

The blue graph show the yaw angle ... and the blue background the active track mode (the black - heading speed needs to be taken times 10 to be correct).
View attachment 113620

Secondly the take off spot & the TXT event log make me suspicious ... can we be dealing with a yaw error due to a magnetic disturbed take off spot kicking in late in the flight due to the low flight path with little need for the FC to hold/correct position?

This is the flight area & the take off spot ... a lot of magnetic thing's every where on ground level with those tram tracks.
View attachment 113622
View attachment 113623

Then we have this in the TXT event log ...
View attachment 113624

In retrospect ... this area wasn't the best place to fly a drone, especially if it was a lot of traffic there.
You've got alot of heavy foilage there as well. Could have been adding and dropping satellites and got confused as to where it was...just a thought
 

slup

Well-Known Member
Premium Pilot
Joined
Oct 18, 2019
Messages
3,768
Reactions
8,926
Age
54
Location
Sweden
You've got alot of heavy foilage there as well. Could have been adding and dropping satellites and got confused as to where it was...just a thought
The Navhealth & the sat count from the log there in the moment was fully sufficient for a stable flight ... so it's unfortunately not as easy as that.
 

Meta4

Well-Known Member
Joined
Oct 30, 2016
Messages
11,368
Solutions
1
Reactions
13,363
Age
66
You've got alot of heavy foilage there as well. Could have been adding and dropping satellites and got confused as to where it was...just a thought
Even if satellites were dropped (they weren't), that doesn't switch the drone to Active Track or make it fly a complicated path.
With no joystick input as is seen in the recorded data, it would only hover or drift on any breeze that might have been acting.
 
  • Like
Reactions: Deleted Member 997

Zoogy

New Member
Joined
Sep 22, 2020
Messages
4
Reactions
0
Age
53
Location
New Orleans, LA
The MCDatFlightRecords folder is empty.

There is another folder called Flight Logs which contains a bunch of .dat files.
 

beeflyin

Well-Known Member
Joined
Sep 21, 2020
Messages
78
Reactions
21
Location
32608
About all anyone can tell from looking at the flight data is that from 3:52.9 you centred the joysticks and left the drone hovering a little above head height
At 4:02.9 the flight mode was changed from P-GPS to Active Track and at 4:05.6 it started to fly backwards to the south (across the road), and then west, parallel to the road, accelerating to 17.7 mph (at 4:17.1) and then beginning to slow down.
At 4:13.4 you started to give joystick input with throttle, rudder and aileron to vary the flight path.
At 4:16.7 you pushed the right stick hard to the left til 4:18.3.
At 4:18.2 there is a warning of an upward obstacle detected.
At 4:18.3 the pitch, roll and yaw data indicates a collision.

It's not clear what the drone was tracking or if it would have avoided the crash on its own.
Your full stick move might have contributed to the collision.

For future reference, the quickest way to cancel any automated flight (like Active Track, RTH etc) is to flick the flight mode switch to Sport and back to P-GPS.

A few weeks ago I was flying my Mavic 2 Pro with a friend. I flew it once or twice in Active Track with my friend walking away from me a short distance, maybe 20 feet, just to show him how it works. Then I exited active track and flew the drone back to my position.

After a few very short flights back and forth between me and my friend (in P-GPS mode), the drone was hovering before me for a few seconds, and then it suddenly took off diagonally to the left and crossed the street, which was full of moving cars. The Mavic barely escaped hitting a moving car and then crashed into a tree. I tried to gain control, pushing up on the stick for altitude and hard left to get it out of the street, but my stick movements did not seem to have much affect on the drone.

I am certain that I did NOT put the drone into Active Track mode before the flyaway.

The gimbal/camera and forward facing sensors need to be replaced.

I purchased the drone just under a year ago, so it is still under warranty and I was hoping that DJI would analyze the logs and offer to cover repairs. They just responded that the drone was operating normally in Active Track mode and therefore they would not cover repair costs.

I am posting the log here, hoping to get some feedback from the forum on what happened. HOW did it go into Active Track mode? And WHAT was it tracking? There is no way that a person, animal or object crossed the street at that time. Why did it suddenly take off, away from me and my friend, and how do I convince DJI that it was NOT "working as designed"?

Here is my log at Phantom Help: DJI Flight Log Viewer - PhantomHelp.com

Thanks for taking a look!

PS - here is the response from DJI tech support:

Dear Daniel,

.
.
.

FLY050

1. Unit was in ACTIVE TRACK mode and was working as designed.
2. At t=461s, relative height=1.7m unit recorded external impact.
3. User commanded ROLL left prior to impact.
4. Map matches crash location with trees. Tree branches, Cables, Light poles, slanted roof, poor lighting. This is not detected by vision system due to hardware limitation.

Force impact date: 2020-08-19
Force impact GPS location: 29.9303359, -90.1170812

According to the analysis, the incident was not caused by any product malfunction.

We can still proceed with the repair, however, you will cover the repair charges.
I had the same thing happen with my Mavic Air 2 last week. I know I never switched it to Active Track, and about 3 min before it crashed, it was in Active Track. Did you ever get it figured out?