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

Obstacle Avoidance

dj_ricecube

Member
Joined
May 30, 2017
Messages
14
Reactions
2
Age
50
Crashed my Mavic this past weekend while in Switzerland. I lost control of it after about 10 seconds from takeoff. It wouldn't respond to any stick inputs. After reviewing the flight records, I had about 8 connected satellites prior to takeoff and then immediately went to 0. So GPS signal was lost as well as a compass error. TBE followed and ultimately crashed into the side of the cliff.

Wouldn't obstacle avoidance come into play if it was coming close to colliding with the cliff wall? Video footage ends with it heading right into the cliff...
 

Attachments

  • DJIFlightRecord_2017-05-28_[16-58-10].txt
    39.6 KB · Views: 7
So the first thing I note is that you're flying with an elevation of 1500 m next to a giant rock face. I'm going to assume that generated quite a bit of wind.

Assuming it was windy, the rock face probably destroyed your GPS reception. I would ideally like to see the DAT file, but it looks to me as if the drone was struggling with the wind and that the wind pushed the Mavic into the wall. Without a GPS lock the drone couldn't maintain position that well.
 
So the first thing I note is that you're flying with an elevation of 1500 m next to a giant rock face. I'm going to assume that generated quite a bit of wind.

Assuming it was windy, the rock face probably destroyed your GPS reception. I would ideally like to see the DAT file, but it looks to me as if the drone was struggling with the wind and that the wind pushed the Mavic into the wall. Without a GPS lock the drone couldn't maintain position that well.

It was actually pretty still that Sunday. I would've welcomed any breeze since it was pretty warm on the 1.5 hour hike up! :)
I can upload the DAT file tonight. Thanks!
 
So the first thing I note is that you're flying with an elevation of 1500 m next to a giant rock face. I'm going to assume that generated quite a bit of wind.

Assuming it was windy, the rock face probably destroyed your GPS reception. I would ideally like to see the DAT file, but it looks to me as if the drone was struggling with the wind and that the wind pushed the Mavic into the wall. Without a GPS lock the drone couldn't maintain position that well.

Dropbox - FLY028.DAT

Dropbox - FLY029.DAT

Dropbox - FLY030.DAT
 
I went over this real quickly and found a few things.

1) You took off with a really poor GPS signal.

2) The craft received your inputs and responded but you just crashed, sorry.

You can see the light purple area at the 11 second mark was when you went into ATTI mode.

The first graph shows your GPS signal health and number of sats

The second graph show your stick inputs

The third graph shows the craft responding to the inputs.

Mountaincrash.jpg

Sorry I did not find anything out of the ordinary, it looks like a straight up loss of control by the pilot due to ATTI mode and wind.

Rob
 
I went over this real quickly and found a few things.

1) You took off with a really poor GPS signal.

2) The craft received your inputs and responded but you just crashed, sorry.

You can see the light purple area at the 11 second mark was when you went into ATTI mode.

The first graph shows your GPS signal health and number of sats

The second graph show your stick inputs

The third graph shows the craft responding to the inputs.

View attachment 14426

Sorry I did not find anything out of the ordinary, it looks like a straight up loss of control by the pilot due to ATTI mode and wind.

Rob
That's unfortunate. Lesson learned, I guess. Thanks for taking the time to look at the files!
 
Crashed my Mavic this past weekend while in Switzerland. I lost control of it after about 10 seconds from takeoff. It wouldn't respond to any stick inputs. After reviewing the flight records, I had about 8 connected satellites prior to takeoff and then immediately went to 0. So GPS signal was lost as well as a compass error. TBE followed and ultimately crashed into the side of the cliff.

Wouldn't obstacle avoidance come into play if it was coming close to colliding with the cliff wall? Video footage ends with it heading right into the cliff...
In addition to the issues pointed out by @Robbyg and @Cyberpower678 this flight was launched from a geomagnetically distorted site. It appears that you did a compass calibration. Was this because the Go App was telling you that it needed to be done? It also appears that after this a compass error was being indicated and that by waiting 50 secs with the Mavic on the ground the compass error went away. Is this right?

After the compass calibration the Mavic was placed on the ground (-111 secs) and the Go App then indicated a compass error.
upload_2017-6-7_6-2-45.png
At -68 the Mavic was picked up and re-oriented. While lifted off the ground the compass error indication went away but then returned when it was placed back on the ground.

Over the next 50 secs the Flight Controller adjusted it's value of Yaw to reflect the erroneous magYaw value. Finally, at -16 the FC had come to accept the erroneous magYaw value and the compass error warning went away. Seeing this the pilot assumed that it was safe to launch.

Note, that when the Mavic achieved 1 meter altitude the magYaw value became correct (the same value of Yaw back at -68 secs right before the Mavic was paced on the ground). But, Yaw was still incorrect. Clearly, the launch site was geomagnetically distorted.

The Mavic's Flight Controller thinks it's smart enough to figure out compass error issues, fix them, and fly anyway. Sometimes that works; usually if the compass error is small. Often times it doesn't work. This flight is an example where it didn't work. The P3 doesn't do this. But, the P3 has the opposite problem in that it will erroneously detect a compass error when one doesn't exist. I suppose the Mavic strategy is DJI's response to those P3 false positive "compass errors". But, how hard can it be attempt to fix just the smaller compass errors and follow the P3 strategy for the larger ones? </rant>

Anyway, since we're stuck with this mis-guided strategy there are two checks that pilots should do before launch.
1. Check the Go App map display and make sure the orientation of the red triangle agrees with the actual orientation of the Mavic.
2. Check the GPS level on the RC and make sure it's 4 or 5. This is not number of satellites. It's the bar graph display.

Doing these two checks would've have prevented this incident as well as several other incidents that I've looked at.
 
That's a very useful post. I'm UK based and took the Mavic to the very South of Portugal.
I received compass errors when trying to take off from a large terrace on the 4th floor - presumably because of steals in the buikding. Eventually found a spot it was happy with and nervously took off - although I wasn't sure how happy it *really* was.

Reading the above I should gave paid more attention to the GPS health bars.
I wasn't sure if the errors were due to:

1-traveling 1,500 miles south (recalibrate?)
2-Building interference
3-huge cell tower just 100m away

Once in flight and away - all was fine and no issues landing.

I elected not to recalculate until I could get into the hills in the middle of Nowhere.
 
That's a very useful post. I'm UK based and took the Mavic to the very South of Portugal.
I received compass errors when trying to take off from a large terrace on the 4th floor - presumably because of steals in the buikding. Eventually found a spot it was happy with and nervously took off - although I wasn't sure how happy it *really* was.

Reading the above I should gave paid more attention to the GPS health bars.
I wasn't sure if the errors were due to:

1-traveling 1,500 miles south (recalibrate?)
2-Building interference
3-huge cell tower just 100m away

Once in flight and away - all was fine and no issues landing.

I elected not to recalculate until I could get into the hills in the middle of Nowhere.
Hard to know for sure but it sounds like your problem was due to some ferrous material distorting the geomagnetic field at the launch site. If so, check 1 would have shown that; i.e. make sure the Go App map display is showing an AC orientation that is the same as what you actually see.
 
Last edited:
In addition to the issues pointed out by @Robbyg and @Cyberpower678 this flight was launched from a geomagnetically distorted site. It appears that you did a compass calibration. Was this because the Go App was telling you that it needed to be done? It also appears that after this a compass error was being indicated and that by waiting 50 secs with the Mavic on the ground the compass error went away. Is this right?

After the compass calibration the Mavic was placed on the ground (-111 secs) and the Go App then indicated a compass error.
View attachment 14489
At -68 the Mavic was picked up and re-oriented. While lifted off the ground the compass error indication went away but then returned when it was placed back on the ground.

Over the next 50 secs the Flight Controller adjusted it's value of Yaw to reflect the erroneous magYaw value. Finally, at -16 the FC had come to accept the erroneous magYaw value and the compass error warning went away. Seeing this the pilot assumed that it was safe to launch.

Note, that when the Mavic achieved 1 meter altitude the magYaw value became correct (the same value of Yaw back at -68 secs right before the Mavic was paced on the ground). But, Yaw was still incorrect. Clearly, the launch site was geomagnetically distorted.

The Mavic's Flight Controller thinks it's smart enough to figure out compass error issues, fix them, and fly anyway. Sometimes that works; usually if the compass error is small. Often times it doesn't work. This flight is an example where it didn't work. The P3 doesn't do this. But, the P3 has the opposite problem in that it will erroneously detect a compass error when one doesn't exist. I suppose the Mavic strategy is DJI's response to those P3 false positive "compass errors". But, how hard can it be attempt to fix just the smaller compass errors and follow the P3 strategy for the larger ones? </rant>

Anyway, since we're stuck with this mis-guided strategy there are two checks that pilots should do before launch.
1. Check the Go App map display and make sure the orientation of the red triangle agrees with the actual orientation of the Mavic.
2. Check the GPS level on the RC and make sure it's 4 or 5. This is not number of satellites. It's the bar graph display.

Doing these two checks would've have prevented this incident as well as several other incidents that I've looked at.

Nice Catch Budwalker.

Rob
 
Thanks so much for that detailed anaylsis and explanation!

That is exactly what happened. I did have to do compass calibration 2x because it told me to. The compass error did go away and I thought it was okay to fly.

In addition to the issues pointed out by @Robbyg and @Cyberpower678 this flight was launched from a geomagnetically distorted site. It appears that you did a compass calibration. Was this because the Go App was telling you that it needed to be done? It also appears that after this a compass error was being indicated and that by waiting 50 secs with the Mavic on the ground the compass error went away. Is this right?

After the compass calibration the Mavic was placed on the ground (-111 secs) and the Go App then indicated a compass error.
View attachment 14489
At -68 the Mavic was picked up and re-oriented. While lifted off the ground the compass error indication went away but then returned when it was placed back on the ground.

Over the next 50 secs the Flight Controller adjusted it's value of Yaw to reflect the erroneous magYaw value. Finally, at -16 the FC had come to accept the erroneous magYaw value and the compass error warning went away. Seeing this the pilot assumed that it was safe to launch.

Note, that when the Mavic achieved 1 meter altitude the magYaw value became correct (the same value of Yaw back at -68 secs right before the Mavic was paced on the ground). But, Yaw was still incorrect. Clearly, the launch site was geomagnetically distorted.

The Mavic's Flight Controller thinks it's smart enough to figure out compass error issues, fix them, and fly anyway. Sometimes that works; usually if the compass error is small. Often times it doesn't work. This flight is an example where it didn't work. The P3 doesn't do this. But, the P3 has the opposite problem in that it will erroneously detect a compass error when one doesn't exist. I suppose the Mavic strategy is DJI's response to those P3 false positive "compass errors". But, how hard can it be attempt to fix just the smaller compass errors and follow the P3 strategy for the larger ones? </rant>

Anyway, since we're stuck with this mis-guided strategy there are two checks that pilots should do before launch.
1. Check the Go App map display and make sure the orientation of the red triangle agrees with the actual orientation of the Mavic.
2. Check the GPS level on the RC and make sure it's 4 or 5. This is not number of satellites. It's the bar graph display.

Doing these two checks would've have prevented this incident as well as several other incidents that I've looked at.
 

DJI Drone Deals

New Threads

Members online

Forum statistics

Threads
130,599
Messages
1,554,238
Members
159,603
Latest member
refrigasketscanada