DJI Mavic, Air and Mini Drones
Friendly, Helpful & Knowledgeable Community
Join Us Now

Mavic 2 crash in active track

dbj702

Active Member
Joined
Feb 9, 2018
Messages
28
Reactions
14
Age
52
Hello Mavic community. Below is the YouTube link to a video of my Mavic 2 Pro crashing in active track. I’m not sure what caused the crash or why none of the sensors prevented the incident. Any input that could explain why this happened is welcomed. Thank goodness this was just a sand dune. The drone suffered no physical or cosmetic damage whatsoever. Please use my experience to prevent any recurrence and to bring awareness to the MavicPilot community of the potential to crash during active track.

Thanks and Enjoy!

 
Sand is very bad news in all things mechanical. Some surely got into your motors & gimbal, so you might be wise to have someone like @Thunderdrones give it a thorough cleaning. Anyway, nice tracking video in some interesting country!
 
  • Like
Reactions: Former Member
It appears that just before the crash, the drone was at a fairly low altitude. The dune buggy then went over a small hill and actually almost seemed to disappear for a moment. It seems that the drone at that point tried to follow it down that small incline.
 
Last edited:
The drone stopped tracking and continued on a track to find it. It was going too fast to find it again.

Why no sensors, no idea. Did the radar go off, was it landing, tough call.
 
  • Like
Reactions: HighlandsVideo
The sand had no contrasting features for the sensors to detect, snow is like this too - had the Mavic had ultrasonic sensors, it wouldnt have done that.

Suggestion: track from a higher elevation, the turns you make will be smoother on video, and it will be less likely to bottom out.
 
The sand had no contrasting features for the sensors to detect, snow is like this too - had the Mavic had ultrasonic sensors, it wouldnt have done that.

Suggestion: track from a higher elevation, the turns you make will be smoother on video, and it will be less likely to bottom out.
Very good info, thanks. I’m pretty new to using active track. This is only my 3rd time using the feature so I’m searching for all the information available. The first time i used active track i was walking, second time i forgot to record, and this time.
 
Last edited:
My guess and it is only based on what I see, I think the tracking algorithm caused the Mavic to lower its altitude incorrectly.
 
Active Track with DJI products has always been poor IMO.

Assuming VPS was functional it should have prevented the crash; although, the sand may have been difficult to get a good reading on.

I've owned many DJI drones with AT 1.0 and heard ÀT 2.0 is better but I'm not seeing it. The M2 has a hard time holding consistent distance from the subject and altitude is all over the place.

I've heard conflicting information on AT 2.0 and some DJI reps don't understand it's operation.

More advanced drones (follow) use a combination of technology for follow that DJI cannot match. It's been frustrating for me since my P3P using DJI for follow, it's a fail.

Halo, Solo, Typhoon......all exceed DJI in follow except in OA. The Parrot Anafi for example uses a combination of optical and GPS tracking for a robust link to the subject even if visual is lost. Then add dynamic altitude control where altitude data from the subject is used to properly set height; the barometer in the tablet is used in conjunction with VPS for proper height control.

There's too many faults with AT 2.0 for hands off follow; I suggest you monitor the flight path at all times and be ready to correct. Wear a helmet with AT 2.0 if not in a vehicle in the event of a AT attack.
 
Maybe it knew its video would be put to that music and killed itself. Mercy. Yes, it lost track in that broad sand plain where the vehicle appears to have sped up. I would also say fly it higher so if it does lose track it can find the vehicle again at a safe altitude.
 
The sensors are confused by sun, and bright sun on featureless sand is probably not something the drone can reliably discern.

Were you in sport mode? Looks like the drone is moving pretty quick - if so that also disables the sensors.

It did a pretty awesome job of tracking the buggy before the crash.

Someone should always have their hands on the controls as well, and if you see it get too low, cancel the active track or hit pause. It's very good, but not infallible, so relying on it blindly is never a good idea.

Do not fly that drone again until you are a million percent sure every single grain of sand is gone from things like the motors, gimbal, camera, etc.
 
Exactly why you need more data for proper follow, sensors are not reliable. GPS and a barometer make a big difference.

 
  • Like
Reactions: drakkor
Wow! Just watched the video! Didn't expect this kind of nosedive!

If we take for granted that sensors were active, then definitely the lack of contrast did not help the bottom cameras to properly & quickly calculate Mavic's height at the time of crash.

The crash happened when the subject was small in the frame. The Mavic tried to move closer to the subject so it could continue tracking it.
It looks to me that the Mavic thought that it had gained altitude which made the subject smaller in frame and it tried to lower in order to be closer to the subject.
During descent the sensors didn't catch the actual height, the movement was very quick and it crashed.
 
Toward the end it seems clear to me that your dune buggy was going too fast for your Mavic to keep up. It lost the tracking lock because the dune buggy was too small (relatively) within the video frame.

Why the Mavic suddenly did a nose dive is beyond me. Your video is 4 minutes long at 2x so that was 8 minutes of flying. Was the battery fully charged when you started?

The video was kind of funny, actually. It was as if the drone just said, "I'm tired and I'm going to sit the rest of this one out!". :)

Mark
 
Looks like the Valley of Fire, Nevada

IMO the AT lost its target as the buggy got too far away and somehow it didn't see the sand dune while searching for the target, as sand dunes can be confusing to Obstacle Avoidance, like water. Just a wild guess though.
 
Lycus Tech Mavic Air 3 Case

DJI Drone Deals

New Threads

Forum statistics

Threads
130,584
Messages
1,554,091
Members
159,585
Latest member
maniac2000