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

Air2 S started landing over water.

Per DJI's documentation, the downward sensors could negatively affect the horizontal position of the drone when flying over moving/reflective surfaces (like in this case).
Yes the optical sensor can be tricked into following moving obects, but it's only used for horizontal position hold when hovering with sticks centred, no? It's never an issue when actively flying around.

I don't know of any advertised VPS functionality that would cause a drone to auto descend into water. However, that data is certainly available and DJI could use it to command the drone (even if they don't advertise they are doing it).
That's what I'm curious about. The VPS infrared height sensors are always (when within range) actively measuring AGL height, as recorded in the flight log. But in all my experiments the drone never reacts in any way to those measurements, unless within 2ft height of an obstacle or ground. Then "Landing Protection" kicks in to automatically raise the drone up to maintain at least 2ft clearance, unless you're holding the throttle stick down to command an auto-landing, after which the height sensors are used to gently land the drone.

If the system is working properly to detect the surface of the water, it can only ever prevent the drone from descending into the water. But if the sensors are not working properly (i.e. registering the bottom of the lake rather than the water's surface), or you've disabled the sensors by sticking gaffer tape over them, then Landing Protection is unable to prevent you from manually lowering the drone into the water.

There is no mechanism by which those VPS sensors can force a descent and ditching into water, unless the pilot actually commands that by holding the throttle down.

Earlier "uncommanded descent" episodes with the original Mini were traced to prop deformation and firmware that prioritized maintaining pitch angle over maintaining height, and that never had anything to do with the VPS sensors.

I'm just really curious where this myth about VPS sensors causing drones to ditch in water came from, and whether anyone has ever demonstrated any conclusive evidence that such a thing is even possible.

re: pressure change of the aerodynamic bow wave from the boat...
That one is outside of my wheelhouse.

Me too. I'm still quite skeptical of that theory.

These guys (Boat Zone) post a lot of YouTube videos from Haulover Inlet in Miami, using their drones to track expensive boats. They frequently fly right up close to some really BIG boats.

See at 8:28, 10:28, and 12:02 in this video.

 
  • Like
Reactions: FrankenCub
Yes the optical sensor can be tricked into following moving obects, but it's only used for horizontal position hold when hovering with sticks centred, no?
Nope. They are also used for precision landing, preventing the drone from landing in an unsafe location, and accidentally descending / crashing into the ground. And maybe more?


I'm just really curious where this myth about VPS sensors causing drones to ditch in water came from, and whether anyone has ever demonstrated any conclusive evidence that such a thing is even possible
Me too. I wasn't aware that such a myth existed.
 
re: pressure change of the aerodynamic bow wave from the boat...

Me too. I'm still quite skeptical of that theory.

Count me as skeptical on that as a cause.

A negative pressure change would be required to cause the drone's barometric sensors to sense a climb and command the drone to descend to compensate. I'm not able to imagine a boat pushing a negative pressure region ahead of it or even over its bow.

It's easy to envision low pressure regions behind a boat or aft of the deckhouse and other structures on deck. There are many examples of drone operations from boats, usually on the after portion of the deck, and I don't recall any reports of sudden descents on departure or landing.
 
Nope. They are also used for precision landing, preventing the drone from landing in an unsafe location, and accidentally descending / crashing into the ground. And maybe more?
Like I said, the VPS height sensors are only used by "Landing Protection" to prevent the drone from crashing in an unintended landing, or are used to ensure a gentle settling whenever an auto-landing is commanded. The sensors cannot trigger a surprise landing for no reason at all.

Me too. I wasn't aware that such a myth existed.

:) And yet, when asked how to prevent the VPS system from possibly causing an unintended descent whenever flying over water, people frequently suggest covering the sensors with gaffer's tape.
Any real way to disable / prevent that from happening in the future? Don't want to have another drop as I frequently fly around water.
There is no setting to disable the downward sensors in DJI Fly. You could cover them with a thin strip of gaffer's tape.

I'm not trying to be difficult about this. I'm just genuinely curious.

My Mini never reacts in any way to changing VPS height measurements, unless within 2ft height when Landing Protection is triggered to prevent it descending closer than 2ft. Any further descent below 2ft is only in response to a command for auto-landing with the throttle stick depressed, or low battery, or automated RTH etc.

If the VPS sensors can somehow trigger and force an uncommanded landing, surely someone could figure out a way to demonstrate this behaviour in an experiment that everyone else could repeat to confirm for themselves.
 
Like I said, the VPS height sensors are only used by "Landing Protection" to prevent the drone from crashing in an unintended landing
"Landing Protection" is not a feature designed for use in unintentional landings. It's a feature designed to help the drone land on suitable ground when intentionally landing.


The sensors cannot trigger a surprise landing for no reason at all.
Is that a myth too? Or are you possibly assuming the drone in this thread initiated a surprise landing? I see no indication in the flight log that it was attempting to land at any point.


And yet, when asked how to prevent the VPS system from possibly causing an unintended descent whenever flying over water, people frequently suggest covering the sensors with gaffer's tape.
I've only seen people suggest using tape to disable VPS since DJI Fly does not offer access to that setting. There are certainly some (less common) flying scenarios where VPS is more of a nuisance than a help, so disabling it can sometimes be beneficial.


If the VPS sensors can somehow trigger and force an uncommanded landing, surely someone could figure out a way to demonstrate this behaviour in an experiment that everyone else could repeat to confirm for themselves.
Again, I'm not convinced the drone was trying to land in this case.

People have documented many cases where the VPS sensors can trigger and force an uncommanded ascent when the drone isn't anywhere near the ground. If that's possible and not mentioned in any DJI manuals, then I don't think other possibilities can be ruled out either just because they are not documented and/or you don't routinely experience it with your Mini.
 
A negative pressure change would be required to cause the drone's barometric sensors to sense a climb and command the drone to descend to compensate. I'm not able to imagine a boat pushing a negative pressure region ahead of it or even over its bow.

It's easy to envision low pressure regions behind a boat or aft of the deckhouse and other structures on deck. There are many examples of drone operations from boats, usually on the after portion of the deck, and I don't recall any reports of sudden descents on departure or landing.

Here's a 15min video by Boat Zone, this one shot entirely from their drones.

Most of the time the drone is flying backward ahead of the boats. But at 11:02 there's this shot from the drone following closely behind a fast moving boat, where there's most likely to be a low pressure zone.
FollowingBehind.jpg

I'd love to see how they do this. Flying backwards while watching the boats means they can't see where the drone is going. I hope they're using a spotter to warn them of oncoming other boats, or jet skis. I'd be worried of the drone getting splashed, or colliding with one of the many other drones similarly operating in this busy area.

OtherDrones.jpg

Nobody ever seems to object to the drones coming so close to their boats, or maybe those people are just never included in the edited videos. But isn't there a Part 107 prohibition on flying over people or over people in moving vehicles?

Even though they get really close, none of the shots actually show the drones flying over any boats though. There is one moment starting at 3:32 where it looks like the boat is actually going to pass under the drone, but there's an awkward jump cut as the video is edited apparently to remove that scene.
 
"Landing Protection" is not a feature designed for use in unintentional landings.
It can prevent unintentional landings.

I demonstrated that in the first of four videos in this playlist.

If you hold the throttle stick slightly lowered, causing the drone to slowly sink, it'll only descend to 2ft (0.5m) above ground no matter how long you hold the throttle partially down. That's because Landing Protection prevents it from going any lower. From there you need to hold the stick fully down [more than 85%] to confirm your intent to engage auto-landing.

It's a feature designed to help the drone land on suitable ground when intentionally landing.
More accurately, Landing Protection is a feature designed to prevent the drone from landing on unsuitable ground.

[...] are you possibly assuming the drone in this thread initiated a surprise landing? I see no indication in the flight log that it was attempting to land at any point.
No, I wasn't suggesting that at all. I was only questioning the advice of covering the VPS sensors with gaffer tape as though that could prevent the VPS from ever causing an unwanted ditching into water.

I've only seen people suggest using tape to disable VPS since DJI Fly does not offer access to that setting. There are certainly some (less common) flying scenarios where VPS is more of a nuisance than a help, so disabling it can sometimes be beneficial.
I mentioned that in my same video. For example, if you're trying to fly through a narrow gap between branches. The VPS height sensor detects a branch closer than 2ft underneath and Landing Protection kicks in to automatically raise the drone, whereupon the drone hits the branch above it. Or you're flying through a small window frame. Same thing, the VPS height sensor sees the window frame closer than 2ft below, then Landing Protection automatically raises the drone, and presto it hits the top of the window frame.

So yes, there are situations where it sure would be nice if DJI allowed the option of disabling Landing Protection. But the DJI Fly app doesn't provide that configuration option. The Litchi app however does provide this option, as shown in my videos.

My only question is why people would think it's a good idea, or even a necessity, to disable the VPS system whenever flying over water as they believe the VPS could somehow cause the drone to drop into the water, when there's no mechanism to explain how that could ever happen.

People have documented many cases where the VPS sensors can trigger and force an uncommanded ascent when the drone isn't anywhere near the ground.
It doesn't have to be near the ground itself. It only needs the VPS height sensors to detect "something" within 2ft of its underside. That could be the top of a bush or tree, or it could be some object suspended and dangling from the drone itself, or even any "solid" enough fog bank.

If the VPS height sensor registers anything within 2ft, Landing Protection automatically causes the drone to rise up.

If that's possible and not mentioned in any DJI manuals, then I don't think other possibilities can be ruled out either just because they are not documented and/or you don't routinely experience it with your Mini.
The DJI manuals are quite vague about the behaviour of VPS, without providing explanation of why it behaves the way it does, merely warning it can be "confused" under certain conditions.

I have once experienced what I thought was an uncommanded auto-landing descent caused by the VPS height sensor on my Mini. But with the help from others on this forum that was diagnosed to have been caused by the VPS infrared height sensor detecting mist coming off the waterfall, and registering that as being closer than 2ft below, even though the drone was still high up above the water's surface.

What confused me was that my flight log data proved the throttle stick was never fully depressed. So what triggered auto-landing to commence from that height?

It turns out the throttle doesn't even need to be fully depressed. Auto-landing is also triggered when all three of these conditions are simultaneously met;
  • VPS height is registering 2ft or less, and
  • throttle is continuously held anywhere below centre for 3 secs or more, and
  • at some point within those 3 secs the throttle dips below 85% closed.
That's exactly what happened. The drone was descending with partially closed throttle for more than three seconds, then encountered mist spraying off the waterfall closer than 2ft below, and the throttle momentarily dipped below 85% closed (never fully closed), when the app suddenly announced "Landing"!

The VPS didn't cause the drone to unexpectedly drop from the sky. My holding the throttle down triggered auto-landing, which is exactly as it is programmed to do under the three required conditions. Luckily I had plenty of time to cancel the auto-landing.

If you're flying only 2ft above a lake's surface, and holding the throttle partiallly down for more than 3 seconds, Landing Protection may be preventing the drone from going any lower. But if you also momentarily dip the throttle more than 85% closed, that combination will trigger auto-landing and you'll have only a brief moment to react before the drone sinks into the water. All you need to do to stop it is raise the throttle, as shown in my first video. But you need to be paying attention to realize what's going on and react accordingly.

Again, it's not the VPS that's causing the drone to sink into the lake. Landing Protection is busy trying to prevent that from happening, not allowing the drone to go lower than 2ft above the water. But if the VPS infrared height sensor signal is being absorbed by the water, rather than reliably reflected back, or if you've stuck gaffer tape over the sensors to prevent them from working, there's nothing preventing you from manually driving the drone into the water.

That's the reason I question why anyone thinks it's a good idea to block the sensors while flying over water? The optical sensor, maybe yes. It can be fooled into tracking moving objects while hovering. But the infrared height sensors cannot create an uncommanded descent.

If I'm wrong about that, I'd really appreciate anyone demonstrating how.
 
  • Like
Reactions: rjwmorrell
My only question is why people would think it's a good idea, or even a necessity, to disable the VPS system whenever flying over water as they believe the VPS could somehow cause the drone to drop into the water, when there's no mechanism to explain how that could ever happen
And which people are you referring to? I don't see anyone in this thread stating the VPS caused the drone to drop into the water.

From watching the video, it's pretty clear the crash into the front of the boat most likely led to that. Had the boat not been there, I suspect the OP would have been able to spare the drone.
 
  • Like
Reactions: RodPad
And which people are you referring to? I don't see anyone in this thread stating the VPS caused the drone to drop into the water.
It's right on the first page of this thread, in post #14 & #15.

The issue could have been VPS related (the downward sensors confused by the water below) as this is where things started to head south:
Yeah I had read about issues like that thought I would need to be lower for something like that to happen. Any real way to disable / prevent that from happening in the future? Don't want to have another drop as I frequently fly around water.

The implication was that VPS being confused by the water could somehow cause things to "head south", or cause "another drop" when flying around water. The impression given was that covering the sensors with gaffer's tape could somehow prevent such a re-occurrence.

I am merely seeking clarification on how the VPS could ever cause a drop into water, and whether anyone could explain or demonstrate the steps by which that could be made to happen.
 
It's right on the first page of this thread, in post #14 & #15.
Ah, so "people" is just me then. You should have just said that :D


The implication was that VPS being confused by the water could somehow cause things to "head south", or cause "another drop" when flying around water.
To clarify, I was attempting to show what seems to be an anomaly in the VPS data at (or very close) to the time of the incident. Was that what caused the issue? I cannot say for sure as I've never encountered an incident like this one in the many flight logs I've reviewed.


The impression given was that covering the sensors with gaffer's tape could somehow prevent such a re-occurrence.
As I've stated many times now, that will only disable VPS. Often when troubleshooting issues to find a cause, it can be helpful to eliminate any potential problems one-by-one.


I am merely seeking clarification on how the VPS could ever cause a drop into water
I think you should start another thread then as you're a tad OT now. This drone didn't drop into the water -- it crashed into a boat.
 
  • Like
Reactions: RodPad
Ah, so "people" is just me then. You should have just said that :D
That would have been impolite and indiscreet of me. :)

But @Jpat, who started this thread, seems to have been left with the impression that disabling the VPS sensors with gaffer's tape would prevent "another drop" when flying around water.
Yeah I had read about issues like that thought I would need to be lower for something like that to happen. Any real way to disable / prevent that from happening in the future? Don't want to have another drop as I frequently fly around water.

Although it didn't apply specifically in this case, that same theory is repeated within many other discussion threads on this forum. I'm just trying to establish whether there is any validity to the theory that VPS can somehow cause a drone to crash into water and that covering the sensors would prevent that.
I think you should start another thread then as you're a tad OT now.
Again, not meaning to be impolite or indiscreet, but it was you who first raised the possibility that this crash could have been VPS related. :)

I do already have another thread going though, with video demonstrations, in case anyone is interested.
mavicpilots.com/threads/yes-landing-protection-can-be-disabled-but.134162
 
Yeah I am a little afraid of getting low above the water again. I love getting low side shots while friends are wake surfing. DJI would not investigate the logs as my drone was out of warranty. They offered me 15% off a replacement air 2s, I decided to just go with the air 3.
 
Although it didn't apply specifically in this case, that same theory is repeated within many other discussion threads on this forum. I'm just trying to establish whether there is any validity to the theory that VPS can somehow cause a drone to crash into water and that covering the sensors would prevent that.
If you're asking if Landing Protection could possibly cause a drone to land in water (in general), then the answer is most definitely yes. All automated processes should be used with great care. One should never fully rely on them.


Again, not meaning to be impolite or indiscreet, but it was you who first raised the possibility that this crash could have been VPS related.
Again, I don't think the crash was caused by the VPS. I did not raise that possibility in this thread. It was most definitely caused by the drone flying into a boat.

As for the VPS, I only stated there appeared to be an anomaly in the VPS data around the time the drone started to descend on its own. Feel free to discuss that point in more detail if you can explain away that anomaly.
 
Yeah I am a little afraid of getting low above the water again. I love getting low side shots while friends are wake surfing.
There's always going to be a risk, but I think you'll be fine if you're actively manning the sticks and watching the drone with your eyes. In this case, it appears you did not try to counter the uncontrolled descent until after the drone had crashed into the front of the boat.

DJI would not investigate the logs as my drone was out of warranty
Well, that's rather disappointing. I don't know if @Zbip57 will be able to sleep tonight now ;)
 
.....
As for the VPS, I only stated there appeared to be an anomaly in the VPS data around the time the drone started to descend on its own. Feel free to discuss that point in more detail if you can explain away that anomaly.
What is anomalous about the VPS data? As per post #34 it would appear that the vpsHeight data becomes valid only after the uncommanded descent starts and had been in effect for 1.5 secs. I.e., the uncommanded descent caused the vpsHeight data to become valid.
 
By becoming valid, are you referring to this value changing to True?

View attachment 166901



What happened here?

View attachment 166902
Up until and including time 4 min 57.4 secs isWaveWork is FALSE so the vpsHeight = 7.4 is meaningless. (The value 7.4 is an artifact that comes from the last time isWaveWork is TRUE at time 4 min 2.2 secs). 0.1 secs later at time 4 min 57.5 secs isWaveWork becomes TRUE making vpsHeight = 3.2 valid.


But... If you're thinking the vpsHeight data should've become valid at a higher altitude. The only thing I can think is that the boat moved under the MA2 when vpsHeight started showing a valid 3.2 meters.
 
But... If you're thinking the vpsHeight data should've become valid at a higher altitude. The only thing I can think is that the boat moved under the MA2 when vpsHeight started showing a valid 3.2 meters.
The VPS data in general seems very erratic. isSwaveWork randomly blips on and off at times and is off many times when the downward sensors should be able to easily see the ground (water) below.

It's tough to know what's going on there since DJI doesn't share the recipe for that secret sauce. I suppose it could be a big nothingburger, but it's common knowledge that the aircraft sometimes uses data from those sensors to automatically ascend/descend. I guess we'll never know what happened for sure since DJI has no interest in exploring the data.
 
  • Like
Reactions: BudWalker
Lycus Tech Mavic Air 3 Case

DJI Drone Deals

New Threads

Forum statistics

Threads
131,132
Messages
1,560,144
Members
160,104
Latest member
Roger-N