Hello,

I've been playing for a few weeks with my Mavic and I love(d) it. I was able to do great shots in the Alps, over rivers, etc. An amazing machine.

Yesterday, I went to a barrage and I wanted to film above water while flying flat. This is barrage so the water was extremely still. I started from the shore (quite a few meters above the water level), adjusted the drone to what I thought was the right altitude (about 1-2 meters above water), then switched to Sport mode (first mistake I guess), and then simply moved forward (right stick up). My drone started moving forward but pretty quickly went down (as if I had pushed the left sticker down) and hit the water. This wasn't far, like 100m away from me.

This is where I panicked, unfortunately, and I guess I might have been able to get it back on track as it bumped 1-2 times (touched, went a bit on the side, up, touched, etc.) and as I panicked, I went down with the left controller (I am incredibly disappointed in myself for making such a stupid mistake... probably a lizard brain reflex that was trying to "push" from the ground, hence go down with the stick...).

Anyway, the drone is lost and I lost some great footage on it, but my desire is now to understand what went wrong. Not at the end, post-hitting the water, as I totally get I handled the situation badly, but before that: why did the Mavic start going down? This is really what initiated this downward spiral and worries me.

I've tried going on web sites that analysis the logs but they didn't show the control activities so I'm missing this for whatever reason. Could somebody help?

I want to know whether....
- There are some specific rules to follow when over water that I missed and led to the drone going down?
- Was the Sport mode associated to another condition the problem? (water? low altitude? other?)
- Did I actually hit the controller and told the Mavic to go down (I'm 99.99% sure I didn't do that but the logs would know)
- Was there some strange and local wind activity that pushed it down?
- Was there some local interference? Would be bizarre as this is in the middle of nowhere (well, still, above a barrage so there must be some electrical field not too far...)

The bottom line is that unless I understand what I really happened, I'll be worried to try anything remotely similar to this in the future and won't be learning anything. For that price, I'd love for this lesson to teach me something valuable

Thanks in advance!

P.S.: I also have the video of the crash as recorded by the iPhone... It is painful for me to watch... but let me know if that would help.

The user MANUAL has specific warnings about flying over glassy surfaces (like water), or surfaces with repeating patterns.  The optical sensors have trouble with those.  The barometer is affected by changes in air temperature that can occur between land and water.  Changes that cause convection air currents for example (winds blowing on shore in the day and out to sea at night).  As a result you fly from warm land (lower airpressure to higher) to cool water.  The barometer commands descend to maintain hover altitude.  My default if something goes wrong is climb.  
See page 25 of Mavice Pro Manual v. 1.6.  You can download the most recent and previous manuals at http://www.dji.com/mavic/info#downloads

Start over water and a higher altitude and then work your way down.

Sorry for your loss.  Hope you were able to recover it.  

It's a case where you really should have turned of VPS. It's a hard lesson and it's easy to panic when you see unusual moves like that, most would have done the same.

Key word is "Sports Mode":  in sports mode OA is turned off and angle of attack is max. at max. possible speed.

I guess You hit right stick up instantly for max. speed which tilts MP at max. angle of attack. Sudden forward movement at max. angle of attack then causes high air-resistance and pressure-differences around the non-aerodynamic MP-body that might influence the barometer, which leads to a flight-path that points downward and which most probably cannot be compensated by (confused?) barometer regarding maintaining precise height within a range of meters. Would be interesting to know where MP-barometer sits to simulate pressure-gradient changes at max. speed in a wind-channel ...................

Maybe if You would have started at 4m height, nothing would have happened.

In winter I did some experiments: I wanted to film a frozen lake with cracked-stapled ice. My position was perpendicular to the flight-path and I started slowly in P-mode with all OA-sensors on with only incremental speed-gain. Visually it kept height adjusted to ~2m nicely but at max.speed (~30km/h) = highest angle of attack suddenly my MP howled and stopped (breaking set to 120%, forward OA-sensors now facing downwards detected "frontal" obstacle = ice)

I adjusted height to ~4m at same setup and it flew nicely without forward-OA-breaking at ~30km/h but visually and on RC I could see that it did not maintain height (went down from 4m to 3m)

When switching to S-Mode and instant throttling to max. I saw my MP diving down instantly and - fortunately in time - let go of the control-stick before my MP smashed into the cracked ice. I dunno if it was by centimeters or millimeters but it recovered without crash.

I never repeated such experiments and learned that close to ground maintaining precise height means meters instead of centimeters ...........

Sorry for Your loss .........

Greetings from Austria

As Angelo said when you put it in SPORT MODE the MP would instantly lean down and forward and since you were too low the inevitable happened.

For your next drone...practice, practice, practice before attempting dangerous stunts.

Sorry dude lesson learned at a high price!

Sport Mode + angle of the MP moving forward + your height off the water + the wind resistance pushing the MP  down + your stick input = disaster.

Total beginner here so be gentle. I've got around 20 take-offs with my MP and I have wondered why you push the left stick forward on UAV's to climb. To me it seems counter intuitive.  Makes more sense to pull the stick back to climb.

Thanks all for your replies.

Bill, yes, I was aware of the documentation regarding "glassy surfaces", but I was expecting this to simply lead to the detection of surface not to be precise/relevant anymore. But since I set the level myself and ordered to "go flat", I thought it would stay as is independently of any faulty surface detection (similarly to a high(er) altitude flight). No, I was not able to recover it. Water is pretty cold there and given it is a barrage and don't know when they operate it, whether they have currents, etc. I decided not to do anything stupid on top of my already stupid act. Based on my calculation, the drone must be 25-30m below surface (at this place, the floor of the reservoir goes down by a 1:2.5 angle. I've reached out to local diving clubs that go there from time to time (it used to be an old village that they put under water, so people go visit some of the ruins) but got no response yet (I could at least get my memory card )

Angelo, thanks a lot for taking the time of the detailed answer. I think it might very well be it. It went fine in P-Mode, at slower speed, but once in S-mode, it directly went downward, which might be a mix of angle of attack, disruption of the barometer and low-level winds that were visible at the surface of the lake.

Dronist, yes, you are obviously right, I felt too confident. In anything, I observe that there is a dangerous gray-zone where you are not a beginner anymore, you start feeling confident, yet, at the same time, you are not mature enough to understand the big picture and the real risks involved. It applies to anything: learning how to drive a car, to ski, etc. and I can clearly see myself straight in the middle of that grey-zone

I went back today to see if the current at possibly pushed the drone to the shore, but that was a (very) long shot and I obviously didn't see anything.

Time to buy a submarin drone? ;)

Thanks again for your detailed responses guys, I really appreciate.

Onward,


sacha

slaboure Posted at 2017-7-23 09:58
Thanks all for your replies.

Bill, yes, I was aware of the documentation regarding "glassy surfaces", but I was expecting this to simply lead to the detection of surface not to be precise/relevant anymore. But since I set the level myself and ordered to "go flat", I thought it would stay as is independently of any faulty surface detection (similarly to a high(er) altitude flight). No, I was not able to recover it. Water is pretty cold there and given it is a barrage and don't know when they operate it, whether they have currents, etc. I decided not to do anything stupid on top of my already stupid act. Based on my calculation, the drone must be 25-30m below surface (at this place, the floor of the reservoir goes down by a 1:2.5 angle. I've reached out to local diving clubs that go there from time to time (it used to be an old village that they put under water, so people go visit some of the ruins) but got no response yet (I could at least get my memory card  )

(oh, and you can see my avatar? This is one of the last part of the video of the drone drowning...)

Sorry for your loss, send an email with the DAT files and TXT files to support.nl@dji.com, the Support will open a case for you and provide you a solution.

DJI Diana Posted at 2017-7-24 00:08
Sorry for your loss, send an email with the DAT files and TXT files to , the Support will open a case for you and provide you a solution.

Will do, thanks Diana.