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

LOSS of CONTROL OVER WATER **EXPLAINED**

Information I’d read said when doing the GPS dance (vertical and horizontal turning) to not perform near metal objects. Is that not correct?
Ah - that has nothing to do with GPS - that's a compass calibration. It's performed to subtract the internal magnetic field of the aircraft from the external magnetic field. If you try to do it in a location with local magnetic interference then it may fail.

Independent of whether the compass was successfully calibrated, taking off in a location with magnetic interference is likely to lead to erratic or uncontrolled flight, not because the home point is inaccurate, but because the aircraft heading is likely to be incorrect after takeoff.

For a more detailed explanation:

A short explanation of compass function, calibration and errors.
 
Information I’d read said when doing the GPS dance (vertical and horizontal turning) to not perform near metal objects. Is that not correct?
A GPS signal is electromagnetic in nature, metal will affect signal strength, but not the actual numbers produced.
 
  • Like
Reactions: 4006448
A GPS signal is electromagnetic in nature, metal will affect signal strength, but not the actual numbers produced.
The only way that metal would affect GPS is if a large piece is placed over the GPS antenna to block the reception of satellite signals.
 
  • Like
Reactions: 4006448
Would your take off point have metal that may not have provided GPS home point accuracy?
The flight data clearly shows that the drone was launched prematurely, before sufficient satellites were acquired.
This had nothing to do with any imaginary interference with GPS.
 
  • Like
Reactions: sar104
This had nothing to do with any imaginary interference with GPS.
Agreed.

The only reason I even mentioned RTH was because I was looking right at the drone at the time and the sound of the waterfall was quite loud, so it took me a moment to even notice the beeping sound. Wondering what that was all about, I glanced down at the screen and didn't take the time to read and digest the red message. I just automatically hit Cancel to stop whatever it was doing because I had just then realized the Mini was descending by itself.

My initial thought at the time was that it had gone into low-battery RTH, and because it was within 20m of the Home location it was just straight down landing instead. That needed to be cancelled right away.

At the 4m 1s time mark the VPS altitude suddenly comes alive reporting 3.6ft height above ground, while it's still 27.9ft above me. At 4m 3.2s, as I'm descending at 25m (25ft) above me, the VPS is suddenly reporting 2ft AGL and then it goes into "Forced Landing".
I just noticed I mis-typed that. It was 25ft, not 25m.

The Mini was still 25ft higher than my vantage point, which itself was about 30ft above the water. But, by the time I eventually got around to cancelling the auto-landing, it had already descended 15ft (4.6m).

I stopped it when it was still 10ft higher than me, or 40ft above the water. So there was no actual danger. Just confusion.

I was bringing the drone back down for a landing at the time, so maybe I actually did have the throttle stick held full down at the time. I could understand how, if the VPS.height is showing only 2ft AGL while the throttle stick is being held full down, that would trigger the auto-land sequence as normal. So maybe it saw a spray of mist that close below it? I dunno.

If there had actually been something that close only 2ft under the drone, as recorded by the VPS height, how did the drone manage to descend 15ft without hitting anything before I got around to cancelling the auto-landing? It's curious.
 
If there had actually been something that close only 2ft under the drone... how did the drone manage to descend 15ft without hitting anything before I got around to cancelling the auto-landing? It's curious.
For the Mini that looks really possible actually ...

Considering the operating range of the VPS sensors something can be 2ft below the sensor but as far as 1,43ft behind it & yet be sensed ... 1,43ft behind will certainly let the Mini pass so it can descend further 15ft. Think it's even possible for the Mini to pass if something is just slightly closer than 1,1ft to the right or left of the AC.

Range.jpg
 
When it says Downward "Vision" System, is that referring to just the optical sensor alone, or do those specified field of view angles apply equally to the infrared height sensors? I'll try to test that later by moving my hand under it while it's hovering, but I suspect the IR sensor beam is narrower.
 
If there had actually been something that close only 2ft under the drone, as recorded by the VPS height, how did the drone manage to descend 15ft without hitting anything before I got around to cancelling the auto-landing? It's curious.
Can any of you log deciphering experts tell me whether or not the throttle was being held full-down at the time that auto-landing was triggered?

I know I was bringing the Mini down at the time. But if the IR sensor truly did detect something as close as 2ft below the Mini, shouldn't the obstacle avoidance feature have prevented further descent?

But if I actually did have the throttle held fully down, when the IR sensor insisted the Mini was only 2ft off the ground, that would have signalled to the flight controller that I was intending to have it land there.

It's at 4min 3.2sec in this log.
DJI Flight Log Viewer - PhantomHelp.com
 
Can any of you log deciphering experts tell me whether or not the throttle was being held full-down at the time that auto-landing was triggered?
Here you go ... definitely a Vertical Low Limit landing initiated by you holding the throttle down ... Throttle value neutral=1024, max ascend=1684 & max descend=364.

1634072657582.png
 
Information I’d read said when doing the GPS dance (vertical and horizontal turning) to not perform near metal objects. Is that not correct?
i believe thats compass calibration dance you refer too..
 
Here you go ... definitely a Vertical Low Limit landing initiated by you holding the throttle down ... Throttle value neutral=1024, max ascend=1684 & max descend=364.
But the throttle was not FULL down. The graph shows throttle at 449 when the Forced Landing kicked in. That's still above the full down value of 364.

You mean to say the throttle doesn't need to be ALL the way fully down to trigger the auto-landing? It was still at about 6%.

The throttle did dip closer to 0% after that, but still not fully zero, and auto-landing was already initiated before that.
 
I dusted off my copy of CsvView upgraded to v4.2.4 and found the data to show when Forced Landing was initiated. Throttle (red) ranges here from +10,000 to -10,000. The Mini's Height (olive) is in metres relative to the benchtop takeoff location. And the VPS Height (broken green line) is in metres.

Here I was already bringing the drone down and back to me. Height was 7.7m (25ft) above the bench I launched from. Throttle was down at -8,742 (not -10,000). And VPS Height for some mysterious reason was 0.5m (1.6ft).

I did pull down even more on the throttle, down to a minimum value of -9,318. But it had already initiated Forced Landing by then.

Landing-01.jpg
 
Here's where I released the throttle stick to reach over and press cancel on the app screen to stop the auto-landing.

Landing-02.jpg
 
And, in panic, I gave it FULL up throttle. Note full scale +10,000 on the throttle.

Landing-03.jpg
 
Bringing it back to me, the IR sensor picks up a ground reflection from the rocks below starting at 6.4m height AGL (21ft), while the Mini is still about 5m (16ft) higher than my bench.

There's a short break in the green VPS height line as I bring the Mini across the railing of the observation platform, then descend to land on my hand. Note FULL down throttle (-10,000) to initiate auto-landing.

Note the VPS height actually dips to zero on my hand, while the OSD height registers about 1m still above the bench height as I briefly flipped the Mini over to check the underside to see if there was anything blocking the VPS sensors.

Landing-04.jpg
 
And here the throttle is zero and the OSD height is zero with the Mini now resting back on the bench.

Landing-05.jpg
 
Here's the screen recording.

It starts out with the Mini flying low over the river 4.7m (15ft) below my bench, but still about 15ft above the water.

 
Yeah - there are a few in this thread confidently professing their ignorance. But you can only ignore so many before the forum is useless.

One of the big problems with the posts here is the sloppy language. For example, when the OP says that the drone can "fly into the water" we don't really know if he means that the drone could descend into the water of its own accord (it can't) or if he means the the operator can unintentionally cause the drone to go into the water without the downward visual sensors stopping him from making the mistake. That's a huge difference, but muddied by sloppy wording.
 
  • Like
Reactions: sar104
But the throttle was not FULL down....
Applying full throttle for a descent isn't needed in order to start the landing ... a couple preconditions needs to be met first though.

1. The VPS height should be 0,5m or lower
2. The throttle should in total be applied for descent longer than 3sec.
3. The throttle should sometime during the 3sec be positioned over a certain threshold which seems to be in the ballpark of a throttle value of -8300 to -8500.

These tests are done with my MA1 ... but I'm sure the Mini1 reacts in the same way.

(Throttle reaches a lowest value of -8394, soon after the landing starts, grey area)
1634113644040.png

(Throttle reaches a lowest value of -8576, soon after the landing starts, grey area)
1634113763364.png
 
  • Like
Reactions: AZDave
Lycus Tech Mavic Air 3 Case

DJI Drone Deals

New Threads

Forum statistics

Threads
130,597
Messages
1,554,229
Members
159,603
Latest member
refrigasketscanada