pro 2 just auto landed on water ,couldnt over ride .gone included csv log please help

[Zbip57]

Where the Green VPS height line reappears at about 118secs, the drone is now low enough for the VPS sensor to detect stuff underneath it.

With no throttle input (Blue line level at mid-point), I'd expect the Red barometric altitude line to remain level, while the Green VPS height line fluctuates as the drone passes over varying terrain height, or as the waves roll past underneath. (We don't know if it's the water level of the wave top itself, or of the significant spray above the wave top.)

Does the Mavic2Pro have the same 0.5m (2ft) Landing Protection feature as my Mini? Note the very first couple of blips in the Green VPS height line between 118-121 secs. The VPS height fluctuates between 0.9ft, up to 2.6ft, down to 1.6ft, back up to 2.6ft, then down to only 0.6ft at the 121 sec mark shown. If my Mini detected a VPS height of less than 2ft, Landing Protection would automatically kick in to raise the drone up.

Note the Blue throttle line stays flat without moving during that period, but the red Barometric height goes up 4ft all by itself from -77.3ft at the start of the Green line up to -73.4ft after those first few squiggles. Then the Green VPS line goes suspiciously flat showing a constant 1.3ft height even thought the throttle is dipped down for a moment.

A couple more tiny squiggles in the green line, then it goes completely flat showing 0.9ft, even though the throttle (Blue line) is abruptly raised for a moment at 135sec. The Red barometric line shows a corresponding 20ft increase in height from -77ft up to -57ft. whereas the Green VPS line suddenly spikes straight up to a flat 9.5ft and sticks there. I suspect here the VPS infrared height sensor already either got splashed by a wave or at least thoroughly misted by the spray, prompted the sudden full up throttle input from the pilot.

Then there are a couple of full down throttle applications to lower the drone right back down 20ft to its previous -77ft nearer the water's surface, while the VPS height remains stuck showing 9.5ft until it suddenly starts working again at 147secs. But it's still not working correctly. The VPS height sensor then, except for a couple of brief moments, sticks at 0.9ft all the way across to the right until about 203secs, where it suddenly comes alive again.



More to follow...

 

[Meta4]

I don't know. It still kinda makes sense.

Look earlier in the flight when the drone is over dry land for clear evidence that the VPS height indication cannot be relied on.

The launch point was 88 ft above sea level and the throttle was used at launch to lift the drone to 7.8 ft.
The drone was flown out with no more throttle until it was 225 ft from home.
At this point the ground level was about 40 ft lower then the launch point or 48 feet below the drone.
But in level flight to this point, the VPS never read more than 6.8 feet and despite the ground falling away, the VPS height came up so it was showing 0.6 ft when the ground level was 48 feet below the drone.

Or later in the flight, when VPS was showing again, when the VPS and baro heights are plotted on the same scale, you see that VPS reading stabilised at 1-2 ft just as it did in the early part of the flight.
This tallies with the OP's comment that: The drone has always had a shifty sensor on the bottom

 

[Meta4]

Sitting on the ground before takeoff, OSD height zero ft, the VPS is showing 0.3ft (~3½ inches). Does the Mavic2 have legs that long?

The 0.3 ft reading is the minimum reading that VPS ever shows for any drone.

 

[Zbip57]

Continuing to the last segment, we pick up where the Green VPS height line comes alive again. It's curious how, when you look at the complete graph, how seldom the pilot ever gives any throttle up input. When the drone is low to the water surface, it's almost always either throttle at midpoint, or more down throttle. Yet the Red barometric height goes both up and down. Is that the Landing Protection automatically trying to lift the drone each time?

At 220secs the Red line shows the drone goes as low as -87.2ft, before a huge full up stick input lifts the drone 116ft all the way up to positive +28.8ft. At the same time the VPS height goes dead again and sticks at an indicated 1.9ft. That's obviously incorrect as the VPS sensor should be well out of range at that height. Again, I suspect the sensor got wet there.

Even during the squiggly bits just before then, the sensor seems to stick momentarily as the Green line shows several completely flat sections, which shouldn't be the case if the drone is going up and down as shown by the red line, or the drone is holding a steady height but as the wave height varies beneath the drone.




Next we'll look at the last little bit, with the scale expanded...

 

[Yorkshire_Pud]

I am a bit puzzled about all the discussion.
Is it now being suggested that the final descent was NOT caused by the combination of the down stick and the VPS reading ( irrespective of whether or not that reading was correct ) triggering the landing because the drone thought it was at, or below, the "landing protection " threshold?
If that is not being suggested then does it matter whether or not the VPS sensor was working correctly, surely all that matters is the value it indicated, isn't it?

 

[Zbip57]

Coming in from the left, the Green VPS height line is still stuck (because it got wet?) displaying 1.9ft while the Red barometric height is coming down and levels off at -79ft.

At about 298secs the Green VPS height sensor line comes alive again. Note again the Blue throttle line. It's only either released at mid-throttle or momentarily depressed. There's no UP throttle applied.

At 301sec, FULL DOWN throttle is applied while the VPS height is showing only 0.9ft. The drone is at -82.6ft height when Auto-Landing is triggered and the throttle is released to mid-point.

From there, the Red line descends steadily another 23ft from -82.6ft down to -105.9ft and splashdown. That means (unless the waves were 23ft high) it's unlikely the VPS height was measured from the wave tops themselves. But the provided photo definitely shows spray extending a lot higher above the wave tops.

My theory is all the previous glitchy VPS readings were ignored, where the Green line was perfectly flat and obviously wrong. But here at the end, the VPS is again accurately measuring off the top of the spray mist. The drone may be 23ft above the water, but the spray is reaching within 2ft of the bottom of the drone.

The peaks in the Green VPS line are consistently about 4 seconds apart, the two big ones indicating only about 6ft clearance, corresponding to wave troughs passing below. The VPS height then sinks to zero at the same time the Red barometric altimeter ends at -105ft when the drone auto-lands into the water.

 

[Meta4]

Continuing to the last segment, we pick up where the Green VPS height line comes alive again.

Just the first 15 seconds of the flight is all you need to see to understand that the VPS sensors are completely unreliable and giving completely unbelievable data.

 

[old man mavic]

no matter how much the flight and subsequent demise is analysed ,it will not change the fact, that regardless of the OPs protestations that it was not pilot error ,unfortunately it was
pilot error doesn't just mean how the aircraft was controlled at the time of the crash ,but there are many other factors in play that ultimately,led to the loss of the drone
the biggest mistake the pilot made was simply taking off with a known faulty VPS sensor,(pilot error)
then compounding that by flying the drone to a location where the faulty VPS sensor could have a bearing on the outcome of the flight.(pilot error)
i am sure all of us are sorry that the OP lost his drone ,he asked for an explanation of the reasons for the crash ,and now he has them