Mavic started descending rapidly on its own?

[greginfinity]

I went for a quick flight today and my Mavic Air seemed to descend rapidly towards the ocean! I was just flying over the edge of the shore when all of sudden I saw it start to descend towards the water on its own. I stopped the MA and then ascended and flew home. I am trying to analyze the logs but cannot find the control stick information on the phantomhelp site. I want to make sure I wasn't doing something dumb.

Attaching log in case anyone can help.

At about 5:55 into the flight, the altitude starts climbing rapidly? But I saw the drone descending rapidly (have footage too. Does this seem like a wacky IMU or something? Thanks for any help!

 

[msinger]

Your downward sensors were detecting an obstacle below your Mavic. That will cause it to auto ascend on its own and switch to forced landing mode when the throttle is moved to the full down position (which is what caused the auto landing).

Did you have any accessories installed below your Mavic (e.g. extended landing gear)? Or were the sensors possibly dirty?

 

[greginfinity]

I don't have any accessories installed. The sensors look clean.

I recall being at full throttle (up positions), and it was moving forward, but descending at the time, not ascending even though the sensors say it was ascending.

 

[greginfinity]

Ok thanks. Just uploaded a video (sorry for garbage quality).

 

[swooshdave]

How far above the water were you before it descended? Many stories about water affecting the downward sensors.

 

[greginfinity]

Maybe 10-15 feet. I know the MP had issues with auto landing over water if you were full stick down descending, but I was simply using the forward stick, not hitting the altitude stick at all. At least to my recollection. I was hoping someone could tell from the log what my sticks were doing when it was descending, or at least tell me how to get that info from the logs.

 

[neggy]

Mavic's and water/reflective surfaces do NOT play well unless you go in and disable sensors.

 

[laurens23]

I had a look at the log and video below my thoughts:

Your video ends when you yaw 180 deg. This is at 365 seconds into the log. your video is 21 seconds long, so starts at 344 seconds.

@344sec: you are cruising at ~4m barometric altitude, or 5.1m ultrasound altitude. you are at full forward elevator, but there is no throttle input.

@348sec: Still at full forward elevator, and no throttle input.There is a small drift in altitude barometric altitude to 4.7m ultrasound altitude still at 5.1m.

@352sec Still at full forward elevator, and no throttle input. There is a small 2 second decent to 3.6m barometric and 3.9m ultrasound altitude.

@354sec You give full power, reduce elevator to 0, and climb to 10.5m barometric and outside ultrasound altitude.

@360 you start to yaw 180 deg as we see in the video.

This all seems very minor. but I agree your video appears to show a bigger decent. so I looked further into the logs:

@344sec you are cruising at 6.6m/s in the next 10 seconds this gradually increases to 9m/s that is a significant increase in forward velocity. The increase in forward velocity gives an impression of flying lower, I suspect that this combined with actually drifting a bit lower resulted in what you experienced.

Your reaction was fast and accurate.

 

[Jameslavis]

I was thinking the air was darn low and those waves would cause some I interference with the sensors. Descent also could be speed pitch and some wind.

 

[msinger]

@greginfinity, I took another look at your flight log and found the data from downward sensors looks fine. You can see the altitude reported by those sensors in the "VPS Altitude" column here:

DJI Flight Log Viewer - PhantomHelp.com

At about 5:55 into the flight, the altitude starts climbing rapidly?

Between 5m 55s to 5m 59s, your Mavic ascended from 12.5 feet to 34 feet. Your log shows you had the throttle stick in the full up position during that time, so that rapid ascend makes sense. Your Mavic stopped ascending after that point as soon as you let go of the throttle stick.

 

[msinger]

I had a look at the log and video below my thoughts:

Your video ends when you yaw 180 deg. This is at 365 seconds into the log. your video is 21 seconds long, so starts at 344 seconds.

@344sec: you are cruising at ~4m barometric altitude, or 5.1m ultrasound altitude. you are at full forward elevator, but there is no throttle input.

@348sec: Still at full forward elevator, and no throttle input.There is a small drift in altitude barometric altitude to 4.7m ultrasound altitude still at 5.1m.

@352sec Still at full forward elevator, and no throttle input. There is a small 2 second decent to 3.6m barometric and 3.9m ultrasound altitude.

@354sec You give full power, reduce elevator to 0, and climb to 10.5m barometric and outside ultrasound altitude.

@360 you start to yaw 180 deg as we see in the video.

This all seems very minor. but I agree your video appears to show a bigger decent. so I looked further into the logs:

@344sec you are cruising at 6.6m/s in the next 10 seconds this gradually increases to 9m/s that is a significant increase in forward velocity. The increase in forward velocity gives an impression of flying lower, I suspect that this combined with actually drifting a bit lower resulted in what you experienced.

Your reaction was fast and accurate.

The video above was actually from about 185 - 204 seconds in the flight log.

 

[laurens23]

The video above was actually from about 185 - 204 seconds in the flight log.

I highly doubt that. The end of the video shows climb first then yaw 180 deg parallel to the beach. At the end of 185-204 he does yaw of 90 degree facing out to the ocean then climb which is not like the video shows.

 

[msinger]

At the end of 185-204 he does yaw of 90 degree facing out to the ocean then climb which is not like the video shows.

Right on. And he then flew backwards toward the shore.

Take another look at that video and the flight log. You'll indeed see it's really from 185 - 204 seconds.

 

[greginfinity]

Interesting, thanks for the detailed analysis! Extremely helpful!

I was definitely flying into a decent wind, so I wonder if you are correct in that the extra forward pitch (and velocity) that happens starting a 344 caused it to descend. I guess it must be part of their algorithms on maintaining inputs vs height etc. In this case, the software could have slowed the drone down to maintain altitude, but for some reason it chose to try to keep up forward velocity and ultimately lost altitude. Is there a way to see if the drone was using full power at this point (if it was unable to maintain altitude because the motors were already maxed out)?

 

[laurens23]

Right on. And he then flew backwards toward the shore.

Take another look at that video and the flight log. You'll indeed see it's really from 185 - 204 seconds.

I still don't see it. Near the end of the video he climbs first then yaw second (almost 180 deg before the video stops), this looks exactly like the movements around 360 sec in the log.

At around 204 sec in the log he does a 90 deg yaw first and climb second. this is not what I see in the video.

 

[msinger]

I still don't see it.

On second thought, me neither. After a 10th look, I see the video does in fact show this section of the flight:



I'm not exactly sure why DJI drones do this, but it's not uncommon for a drone to descend a little bit when flying at full stick (or near full stick in this case) close to the ground.

 

[BaNZ]

So have we conclude that it might be the wind? Based on the video, waves and cloud looks like it is a windy condition.

 

[greginfinity]

Not sure we have a definitive conclusion honestly, but just a theory that the forward pitch + head wind + low flying + DJI algorithm made for a sketchy situation.

 

[Heshamp]

I went for a quick flight today and my Mavic Air seemed to descend rapidly towards the ocean! I was just flying over the edge of the shore when all of sudden I saw it start to descend towards the water on its own. I stopped the MA and then ascended and flew home. I am trying to analyze the logs but cannot find the control stick information on the phantomhelp site. I want to make sure I wasn't doing something dumb.

Attaching log in case anyone can help.

At about 5:55 into the flight, the altitude starts climbing rapidly? But I saw the drone descending rapidly (have footage too. Does this seem like a wacky IMU or something? Thanks for any help!

my bru
this is your problem
Apr 17th, 2018 12:59PM | General / Overview | Drone Flight Log from DJI GO app, version 4.2.6 on Android | Total Mileage: 4,321 ft | United States | Airdata UAV
i analysed your data log file by ( AIRDATA.COM) already i have account there
I found that the wind was 20 km gusting 42 km it was very high
Also you had a very high peach angle out of the drone limitation because that you had a lot of forward and backward sensor
See the attached link file of your flight
Apr 17th, 2018 12:59PM | General / Overview | Drone Flight Log from DJI GO app, version 4.2.6 on Android | Total Mileage: 4,321 ft | United States | Airdata UAV

 

[Heshamp]

My bru
The link i sent to you is a dynamic web site so you can see a lot of information about your drone and buttery as well by taping on the the buttons left and ap so scroll down to see all the peg information
I reviewed all, so your drone and battery is very good , the probem was out of limitation wind speed
Please before any flight check the wind carefull
You can use this program ( UAV forecast ) it is very good for gps kp forecst as well
Thank you