"Over Water" mode for drones

[Meta4]

From the many tests we have done landing on the water, the Mini lands the fastest with very little fight , and I think that adds to how many end up in the water.

All the other drones put up a struggle before we can force it to land.

There's a world of difference between inadvertently flying into the water and a mystery effect that causes a drone to descend into water.

 

[AeroForce1]

I've spent this past weekend reading the tragic stories of Drone splashdowns, and wonder if DJI could develop a "Smart mode" to disable, or adapt the vertical sensors to deal with water issues. Yes, I'm aware that some (but not all) DJI AC allow this by a menu click. I'm thinking, a mode that figures out when the "ground signature" changes to water... and a notification pops up recommending switching to water mode. Now maybe this has been discussed before, and the down sensors are not up to the task... but with the brilliant coders at DJI (and on the balance, they are), there should be a way to reduce these tragic losses experienced by both new and experienced pilots alike.

While I'm not sure if DJI monitors these forums or not. Perhaps some of the more seasoned members might have a way of getting this idea through to engineering at DJI.

Just a thought.

I lost my Mavic 2 Pro over the water in Maine. I hired a diver to go down and get it so I could send it back to DJI to get a new one because I had their 'refresh'. Long story short... DJI basically told me to GO POUND SAND. I've since purchased the SKYDIO 2 drone and their sensors have no problem over water. I've lost my faith in DJI and I own 2 other DJI drones. My 'Go To' drone now is the SKYDIO 2.

 

[JamieB]

My gut is that some of these "sensor problems" over water aren't actually sensor related, but are more related to convective currents. Since bodies of water are usually colder than the surrounding land, you'll often get pretty significant downdrafts which can cause aircraft (unmanned or otherwise) to lose altitude pretty quickly.

 

[Zbip57]

When I see so much consistency in these reports, I believe there maybe something to look into here.

Actually, there has been an enormous amount of inconsistency in these reports. There is so much misinformation out there, that it's difficult to get any accurate diagnosis.

As you can see, there's even inconsistency within just this single thread.

Everyone has an opinion or their own pet theory. The best thing to do is search through as many relevant posts as you can find. Sort through those which make sense to you, versus what sounds outlandish. Then devise your own experiments conducted in a safe location to test and verify your results.

Knowing how and why your drone behaves as it does when subjected to certain inputs, and familiarizing yourself with the proper responses required to safely control the situation will help you to react calmly and efficiently when something unexpected occurs. Catatonic panic is rarely the correct solution. ?

 

[AZDave]

Most body of waters to include lakes is already noted and located on the drones onboard GPS mapping, therefore it would not require as complex programming to combine that with tthe downward looking cameras and sensors.

Fly Safe - Mike

Except that GPS doesn't even come close to having the elevation accuracy needed to do that if you're flying fairly low over water, and that's even if the onboard database was large enough (which it isn't) to hold the data to tell the drone whether it was over water or land.

 

[AZDave]

My gut is that some of these "sensor problems" over water aren't actually sensor related, but are more related to convective currents. Since bodies of water are usually colder than the surrounding land, you'll often get pretty significant downdrafts which can cause aircraft (unmanned or otherwise) to lose altitude pretty quickly.

I doubt that. My Mavic Air 2 responds remarkable fast to convective air currents. I often fly in a steep canyon near me house and I've had the drone just hover to check exactly that, and it is remarkably stable. I guarantee that the wind swirls in that canyon are stronger by far than any convective current caused by temperature.

More likely the water causes a change in the barometric pressure of the air over the water, since water will CHANGE the temperature of the air above it much faster than the land will change the temperature of the air above it. Depending upon the location and time of year, it's entirely possible that the water is actually warmer than the surrounding land if somebody is flying late in the afternoon or early in the morning. The land cools much faster than the water once the sun is gone.

Water that's warmer than the takeoff point on land would make the air above it less dense, making the drone think that it was higher than it really was. Some of these people are flying just a few feet off the deck, and that's well within the possible error of likely difference in air density between the air over water and land.

 

[Meta4]

I lost my Mavic 2 Pro over the water in Maine. I hired a diver to go down and get it so I could send it back to DJI to get a new one because I had their 'refresh'. Long story short... DJI basically told me to GO POUND SAND.

That's interesting, but what caused the incident?
Did you have the flight data analysed to find out?

I've since purchased the SKYDIO 2 drone and their sensors have no problem over water.

If so, that would be just like DJI drones as their sensors also have no problem over water.

 

[Guido Cangelosi]

Good intentions, but probably not cost-effective. The truth is, I fly over water all the time without any issues...but not at low altitudes. As log as you recognize certain environments/terrains (like flying over water or flying under bridges) pose additional risk, you can take the necessary precautions for a safe RTH.

What altitude do you consider safe when flying over water ?

 

[Guido Cangelosi]

What altitude do you consider safe when flying over water ?

Cost effective ? well my latest game computer with extra fast CPU and GPU with SSD drives all over the place costed me less than a M2Pro - so why can't we get faster chips there too ?

 

[LA 2020]

I've spent this past weekend reading the tragic stories of Drone splashdowns, and wonder if DJI could develop a "Smart mode" to disable, or adapt the vertical sensors to deal with water issues. Yes, I'm aware that some (but not all) DJI AC allow this by a menu click. I'm thinking, a mode that figures out when the "ground signature" changes to water... and a notification pops up recommending switching to water mode. Now maybe this has been discussed before, and the down sensors are not up to the task... but with the brilliant coders at DJI (and on the balance, they are), there should be a way to reduce these tragic losses experienced by both new and experienced pilots alike.

While I'm not sure if DJI monitors these forums or not. Perhaps some of the more seasoned members might have a way of getting this idea through to engineering at DJI.

Just a thought.

A lot of the problem is sensors get tricked with reflections over water. It is best just not to fly too low. I stay at least 50' over water and on an exceptional sunny day even higher

 

[StantonZ]

What altitude do you consider safe when flying over water ?

Minimum of 50 feet...and typically 100 feet. I generally fly between 100-200 feet over all terrain.

 

[jkissing]

We have a pretty good idea by now how DJI drones behave.
There are thousands of flyers and none have experienced the alleged over-water effects.

While hovering about 15 ft above a moderately flowing river the air 2 started to wabble uncontrollably. I landed and tried later with the same scarry kind of wabble.

 

[baddog]

I've spent this past weekend reading the tragic stories of Drone splashdowns, and wonder if DJI could develop a "Smart mode" to disable, or adapt the vertical sensors to deal with water issues. Yes, I'm aware that some (but not all) DJI AC allow this by a menu click. I'm thinking, a mode that figures out when the "ground signature" changes to water... and a notification pops up recommending switching to water mode. Now maybe this has been discussed before, and the down sensors are not up to the task... but with the brilliant coders at DJI (and on the balance, they are), there should be a way to reduce these tragic losses experienced by both new and experienced pilots alike.

While I'm not sure if DJI monitors these forums or not. Perhaps some of the more seasoned members might have a way of getting this idea through to engineering at DJI.

Just a thought.

Obstacle avoidance is a nice feature and adds to situational awareness. But, was never intended to be 100% reliable, at all times. Safe way to fly over water, keep your drone in VLOS at all times. I try to fly no lower than 8-10 ft above water or the ground and I know where my drone is and what its attitude is. The ground is worse than water IMHO, ground level can rise quickly, at a moments notice.

 

[Meta4]

A lot of the problem is sensors get tricked with reflections over water. It is best just not to fly too low. I stay at least 50' over water and on an exceptional sunny day even higher

That problem is just difficulty with horizontal position holding because the water surface is unsuitable for the VPS sensors. But that doesn't affect vertical position holding which is managed by a barometric sensor.

 

[Scrufdog]

I do most of my flying over water, so I see the slow dive to death into the water all the time. It's much worse on Sport, almost to the point of a dive. If I hover 10ft or less over water, and just sit there, in any flight mode, within about 5 seconds my MA2 with slowly creep down until I tap upwards. Almost lost her a couple times in the beginning due to that, now I expect it.

 

[Alex 007]

Those were caused by prop deformation, totally unrelated to water. Previous firmware versions warned of "Max Power Reached" errors. Firmware 1.0.5 introduced a new "Motor Speed Error" warning with explicit instructions to replace the propeller blades on the affected motor identified by the beeping ESC after landing. The warning was later revised to even more explicitly state, "Propeller rotating too fast", again with instructions to replace the propeller blades identified by the beeping ESC.

In particular the rear prop blades are prone to deformation when stored folded crossed laterally. The prop deformation (twist) reduces their lift capability thereby forcing the motors to spin faster to generate the required lift. Eventually the rear motors are incapable of spinning any faster and can't maintain the required tail-high nose-down pitch attitude in forward flight,

Someone please correct me if I'm wrong, but this is what I understood to be the cause of "Uncommanded Descents". The firmware defect was that the flight controller prioritized the commanded pitch angle rather than altitude hold. Full forward control stick commanded the flight controller to hold a certain nose-down pitch angle to make the aircraft fly forward. If the rear motors were incapable of providing the required lift to achieve that pitch angle, the flight controller responded by decreasing the front motor speeds, thereby maintaining the correct pitch angle. But, with the rear motors already struggling, and the front motors now commanded to spin slower, despite holding the required pitch angle for forward flight the overall lift was now insufficient to hold altitude.

Feeding in more throttle would not stop the gradual descent. If the aircraft was over land or snow, it might just bounce off the ground. That may correct the pitch angle back to level allowing the aircraft to then go back to flying "normally" until the same conditions are eventually repeated. But if the aircraft descends into a lake, well that's terminal.

If you ever do encounter such a situation, others have noted that merely halting forward flight, even pulling full reverse stick, can stop the uncommanded descent and save your drone.


Actually, there has been an enormous amount of inconsistency in these reports. There is so much misinformation out there, that it's difficult to get any accurate diagnosis.

I don't know if the firmware was ever actually "fixed", beyond introducing ever more dire error warning messages. You'll find hundreds of forum posts from people insisting the newly introduced Motor Speed Error warning in firmware v1.0.5 is itself a firmware bug. They had never seen this warning before, it only appeared after they upgraded to v1.0.5. Some even recommended the "cure" of downgrading to a previous firmware version to make the annoying warning go away. Brilliant.

In any case, the uncommanded descent has nothing to do with the downward facing sensors and is not caused by flying over water. Take care of your propeller blades. If you suspect any issue with the blades, do the hover test and plot motor speeds to see what's really happening.
mavicpilots.com/threads/motor-speed-error-mavic-mini.86130/page-13#post-1038918

That said, the downward sensors can be confused by water, but they will not drive the drone down into the water. The vision sensor camera fixes on any patterned surface detail to very accurately hold position in a hover. If the surface is blank with no detectable pattern, or it's just too dark, then that sensor won't work. Or, if the detected pattern is moving, like waves or ripples on the surface of water, the sensor will follow those. Try hovering over a patterned rug and dragging the rug away. The drone will follow.

The barometric altimeter is usually quite accurate, but is affected by changing air temperature and local air pressure. The infrared height sensors (when within range) bounce a signal off the surface to accurately measure altitude. If anything gets within (~1m?) of the bottom of the drone, the sensor actually causes the drone to rise up. Hold your hand under it and you'll see. It'll never cause the drone to descend. To force it to land you need to hold full down throttle.

The last sentence of your comment hits the nail on the head. It is actually the built in barometer that maintains the altitude that the drone is flying at. The GPS really only maintains latitude and longitude and it not accurate for altitude hold at all. The problem arises when there is a sudden change in barometric pressure and the drone can suddenly descend or ascend by a couple of meters. Now normally over land the downward sensors will kick in and avoid a mishap. However over a smooth surface like water we all know that they do not work.

I have a toy Bugs3H drone that only has a barometer as a flying aid that I bought to learn to fly with FPV goggles and this happens very often. It maintains altitude, or altitude hold as they call it very well but can suddenly rise up or down at times when the weather is changing.

 

[AZDave]

That problem is just difficulty with horizontal position holding because the water surface is unsuitable for the VPS sensors. But that doesn't affect vertical position holding which is managed by a barometric sensor.

Except that the barometric pressure over water is not likely to be the same as it is over nearby land where the drone took off. The water is typically warmer or cooler than the surrounding land, and therefore so is the air imediately above them .... and therefore so is the air density. Flying down low over water is simply a bad idea. The baromteric pressure could easily be different than expected and as you say, the bottom facing visual collision avoidance sensors are not reliable over water.

People keep doing it, though, which means we keep seeing these tales of woe, most of which blame some mysterious failure of the drone.

 

[Meta4]

Except that the barometric pressure over water is not likely to be the same as it is over nearby land where the drone took off.

I would be very surprised if there's any significant issue with air pressure.
If there was any difference, it would be right on the land/water edge and even then it would probably be trivial.
Air pressure certainly wouldn't account for the imagined issues flying over water.

Almost all my flying is over water and I think it's safer flying over water than on land.
There are a lot less obstacles to worry about.

 

[AZDave]

I would be very surprised if there's any significant issue with air pressure.
If there was any difference, it would be right on the land/water edge and even then it would probably be trivial.
Air pressure certainly wouldn't account for the imagined issues flying over water.

Almost all my flying is over water and I think it's safer flying over water than on land.
There are a lot less obstacles to worry about.

No ... it won't be most significant near the water's edge. It will have more chance to be different the further from shore you go. That's simple physics. The barometric basis is set at takeoff, and it's not like some sort of sharp transition at the shoreline is the problem. That's probably minor. The problem is the difference in air density that almost certainly gets greater as you go further out over the water.

I have no problem with people flying over water ... as you say it's safer than over land (fewer obstructions) as long as you maintain a significant height above the water to account for air pressure changes. The folks I think are terribly misguided are the ones who think it's cool to cruise two feet off the deck at speed. We're going to keep hearing from their misfortunes.

 

[Zebra48]

There's a lot of confusion about water and sensors.
I haven't seen a splashdown incident that could be blamed on the sensors.
The sensors don't pull drones downwards into water and you shouldn't have any problem flying over water whether the sensors are on or off.

The sensors can be confused by water.
It won't allow horizontal position holding, just like featureless solid surfaces won't.



They don't (and they barely monitor their own forum).

I've flown my MP2 over water, mostly lochs (lakes) and burns (creeks) here in Scotland without any problems. Over the lochs I was between 10ft and 80ft above the water surface at different stages of different flights. Above the burns I was flying at about 6ft above the water level and the water level was at the foot of an embankment about 4ft below the take-off place. The only problem that I had was trying to fly above the burn and then go under a single arch bridge. As I reached the entry, the proximity anti-collision sensors took over and the drone stopped going forward and remained in hover.