Height limitation flying over water

[MavicFlyer2]

Here's a shot from when I was a bit further out and much higher, but you can see conditions.

You sure have that scene framed in very nice, even with a variety of objects for my eyes to take in...Nice Photo and thank you for sharing !

 

[FLDave]

The manual says:

The Downward Vision System and Infrared Sensing System helps the aircraft maintain its current​

position, hover in place more precisely, and to fly indoors or in other environments where GNSS is​

unavailable.​

That's discussing horizontal position holding, not vertical.

So how does the downward infrared system measure horizontal distance and not the vert6ical component?

 

[Meta4]

So how does the downward infrared system measure horizontal distance and not the vert6ical component?

The infrared only measures vertical distance

 

[FLDave]

The infrared only measures vertical distance

Of course it does; I never stated otherwise.

 

[Meta4]

Of course it does; I never stated otherwise.

Your post reads like you did.

 

[Zbip57]

According to the Air 2s manual (page16) the Downward Vision System and Infrared Sensing System helps the aircraft maintain its current position [that would be vertical positioning].
Page 18 gives various conditions when the vision system may not work as expected.

Nope. Nope. No.

It's important to realize that there are two components to the VPS system. There is an optical sensor, and an infrared sensor.

The infrared sensor bounces a signal off of whatever is below to measure the drone's current height. But, (anyone who knows better, please correct me if I'm wrong), this sensor only ever comes into play to actively adjust the drone's height if the sensor detects something within (~2ft?) below the drone. Anywhere above that height, the VPS infrared sensor plays no active role whatsoever in controlling the drone's height.

For example, with the drone hovering in front of you, hold your hand out close underneath the drone and the drone will automatically ascend away from your hand until it reaches (~2ft?). It's like an object avoidance sensor. If you're flying slowly close to the ground, the drone will never descend closer to the ground than that, unless you hold the throttle stick down to confirm your intention to land.

There are obvious limitations to how well this sensor works. If the drone is flying too fast, it cannot react quickly enough to avoid ground obstacles. If it's flying over surfaces that absorb the infrared signal (like water), or scatter it so not enough of the signal is reflected back to the drone, then it won't work.

Many people will tell you that you should always disable VPS whenever flying over water as it (supposedly) can cause the drone to fall into the water. That's nonsense. It's just that the infrared sensor may not work effectively to prevent you from flying too low and driving the drone into the water.

Then there's the other component of the VPS system, the optical sensor. It is used for horizontal position holding, and only if the primary system of the GPS sensor isn't working. If you're flying indoors, in a tunnel, or in a deep canyon, or between tall buildings, with too much of the sky blocked from view of the GPS antenna, then insufficient GPS reception hinders the drone's ability to hold position while hovering with control sticks centred. That's when the VPS optical sensor comes into play.

But, again, there are obvious limitations to how well the optical sensor functions. If it's too dark, of if the surface that it "sees" is completely featureless, like a field of uniformly white snow, or the glassy smooth surface of water, or any other uniformly coloured floor, etc, then the optical sensor sees nothing to lock on to and it will be unable to hold the drone in one position whenever the control sticks are centred.

There are other situations that might result in unexpected consequences. The optical sensor may lock on to something that's moving, like wind ripples or waves on water. I've had it happen to me. Flying in a narrow river canyon, the drone lost GPS signal and switched instead to relying on its VPS optical sensor to hold position when hovering. It was hovering over the river at the time and the optical sensor locked onto some of the foam bubbles floating on the water's surface. With the control sticks centred, I expected the drone to remain hovering in one spot. Instead it started drifting downstream all by itself tracking those moving foam bubbles.

If both GPS and VPS are not working, the drone will switch to Atti mode. In that event, altitude is still automatically controlled, as always, by the barometric altimeter. But the drone will be unable to maintain horizontal position hold and will freely drift with the wind whenever the control sticks are centred. You can still control the drone's movement as normal using stick inputs, it just won't brake to a halt by itself or hold position in a hover.

 

[FLDave]

Nope. Nope. No.

It's important to realize that there are two components to the VPS system. There is an optical sensor, and an infrared sensor.

The infrared sensor bounces a signal off of whatever is below to measure the drone's current height. But, (anyone who knows better, please correct me if I'm wrong), this sensor only ever comes into play to actively adjust the drone's height if the sensor detects something within (~2ft?) below the drone. Anywhere above that height, the VPS infrared sensor plays no active role whatsoever in controlling the drone's height.

For example, with the drone hovering in front of you, hold your hand out close underneath the drone and the drone will automatically ascend away from your hand until it reaches (~2ft?). It's like an object avoidance sensor. If you're flying slowly close to the ground, the drone will never descend closer to the ground than that, unless you hold the throttle stick down to confirm your intention to land.

There are obvious limitations to how well this sensor works. If the drone is flying too fast, it cannot react quickly enough to avoid ground obstacles. If it's flying over surfaces that absorb the infrared signal (like water), or scatter it so not enough of the signal is reflected back to the drone, then it won't work.

Many people will tell you that you should always disable VPS whenever flying over water as it (supposedly) can cause the drone to fall into the water. That's nonsense. It's just that the infrared sensor may not work effectively to prevent you from flying too low and driving the drone into the water.

Then there's the other component of the VPS system, the optical sensor. It is used for horizontal position holding, and only if the primary system of the GPS sensor isn't working. If you're flying indoors, in a tunnel, or in a deep canyon, or between tall buildings, with too much of the sky blocked from view of the GPS antenna, then insufficient GPS reception hinders the drone's ability to hold position while hovering with control sticks centred. That's when the VPS optical sensor comes into play.

But, again, there are obvious limitations to how well the optical sensor functions. If it's too dark, of if the surface that it "sees" is completely featureless, like a field of uniformly white snow, or the glassy smooth surface of water, or any other uniformly coloured floor, etc, then the optical sensor sees nothing to lock on to and it will be unable to hold the drone in one position whenever the control sticks are centred.

There are other situations that might result in unexpected consequences. The optical sensor may lock on to something that's moving, like wind ripples or waves on water. I've had it happen to me. Flying in a narrow river canyon, the drone lost GPS signal and switched instead to relying on its VPS optical sensor to hold position when hovering. It was hovering over the river at the time and the optical sensor locked onto some of the foam bubbles floating on the water's surface. With the control sticks centred, I expected the drone to remain hovering in one spot. Instead it started drifting downstream all by itself tracking those moving foam bubbles.

If both GPS and VPS are not working, the drone will switch to Atti mode. In that event, altitude is still automatically controlled, as always, by the barometric altimeter. But the drone will be unable to maintain horizontal position hold and will freely drift with the wind whenever the control sticks are centred. You can still control the drone's movement as normal using stick inputs, it just won't brake to a halt by itself or hold position in a hover.

Of course there are 2 components to the VPS system; I never stated otherwise.

 

[Zbip57]

The infrared only measures vertical distance

Of course it does; I never stated otherwise.

I'm not trying to turn this into He said, She said scenario. But, specifically, you said:

According to the Air 2s manual (page16) the Downward Vision System and Infrared Sensing System helps the aircraft maintain its current position [that would be vertical positioning].
Page 18 gives various conditions when the vision system may not work as expected.

The VPS system is not used to control vertical positioning, unless there is something closer than 2ft underneath the drone.

The VPS is a vertical positioning reference and the reflection confused the VPS system so that it has no clue about vertical positioning. Therefore a failsafe minimum altitude is established as 30 feet ATL. The drone knows that it is safe to descend to that level but no more as the VPS isn't working over the mirror. When over land VPS can sense something, even if not precise over something like 60 feet, but the mirror below is a vertical positioning error or confusion so the last safe altitude becomes failsafe.

If we're trying to come up with a reasonable explanation of why Jman13's drone wouldn't respond when commanded to descend below 30ft, then your hypothesis is a total red herring. The VPS system just doesn't operate that way.

The only way the VPS system might have been involved is if the drone was trying to descend into a fog layer. If the infrared sensor detected that as an obstacle closer than 2ft below it, then I might prevent the drone from descending any further. But holding the throttle stick full down under those conditions would trigger it to commence auto-landing. The drone would then descend slowly until contacting any hard surface that physically stops it going any lower (i.e. ground, treetop, whatever). Once started, auto-landing can be interrupted to prevent it "landing" in the river.

 

[FLDave]

I'm not trying to turn this into He said, She said scenario. But, specifically, you said:


The VPS system is not used to control vertical positioning, unless there is something closer than 2ft underneath the drone.


If we're trying to come up with a reasonable explanation of why Jman13's drone wouldn't respond when commanded to descend below 30ft, then your hypothesis is a total red herring. The VPS system just doesn't operate that way.

The only way the VPS system might have been involved is if the drone was trying to descend into a fog layer. If the infrared sensor detected that as an obstacle closer than 2ft below it, then I might prevent the drone from descending any further. But holding the throttle stick full down under those conditions would trigger it to commence auto-landing. The drone would then descend slowly until contacting any hard surface that physically stops it going any lower (i.e. ground, treetop, whatever). Once started, auto-landing can be interrupted to prevent it "landing" in the river.

The infrared portion of the VPS is used for a lot more distance than 2 feet. The VPS is composed of the camera part and the infrared part. The infrared part does measure vertical distance.
If my hypothesis is wrong then prove it with flight data.

 

[Zbip57]

The infrared portion of the VPS is used for a lot more distance than 2 feet. The VPS is composed of the camera part and the infrared part. The infrared part does measure vertical distance.

It is measured (and logged), but not used.

If my hypothesis is wrong then prove it with flight data.

That's not how this works. It is your responsibility to present data to support your own hypothesis.

I have already shown in post#15, using the available flight data, that the drone's height does not change when the throttle stick is centred during the time it was flying forward from its takeoff location out over the river. That is because the drone's height is measured and controlled by the barometric altimeter.

While the VPS infrared sensor does measure and record the fluctuating height above ground as the drone moved over the downslope to the river's edge and passed over a tree, it in no way affected the drone's height, which remained constant the entire time because it is controlled by the barometric altimeter, not the VPS sensor.

VPS is not used for vertical positioning. The barometric altimeter does that. You can see that quite clearly in the Phantomhelp log viewer. Look at the IMU Altitude numbers compared to the VPS Altitude starting from the moment of takeoff at the Home Position.
...

The infrared height sensor of the VPS sensor plays no active role in controlling the drone's height unless it detects something within (2ft?) of its underside.

 

[FLDave]

It is measured (and logged), but not used.


That's not how this works. It is your responsibility to present data to support your own hypothesis.

I have already shown in post#15, using the available flight data, that the drone's height does not change when the throttle stick is centred during the time it was flying forward from its takeoff location out over the river. That is because the drone's height is measured and controlled by the barometric altimeter.

While the VPS infrared sensor does measure and record the fluctuating height above ground as the drone moved over the downslope to the river's edge and passed over a tree, it in no way affected the drone's height, which remained constant the entire time because it is controlled by the barometric altimeter, not the VPS sensor.


The infrared height sensor of the VPS sensor plays no active role in controlling the drone's height unless it detects something within (2ft?) of its underside.

I never stated anything to dispute the flight logs. The barometric data is well understood as well as the infrared measurement. I made no inference that the VPS measurement managed or controlled flight as is passed over land.
My comments were to provide a possible explanation of the failure to descend below about 30 ft ATL. Do you have a more plausible explanation for that?

 

[Zbip57]

I made no inference that the VPS measurement managed or controlled flight as it passed over land.

But you suggested that, upon losing VPS height measurement over the water, it triggered some sort of failsafe minimum altitude of 30ft.

When it first lost total VPS reference the ATL altitude was 30 feet. The VPS is a vertical positioning reference and the reflection confused the VPS system so that it has no clue about vertical positioning. Therefore a failsafe minimum altitude is established as 30 feet ATL. The drone knows that it is safe to descend to that level but no more as the VPS isn't working over the mirror.

Clearly that makes no sense at all, because vertical positioning is controlled by the barometric altimeter, not the VPS sensor.

It makes no difference why it loses the VPS height reference, whether it's over land or water. It will also lose its VPS height reference over land whenever the drone climbs high enough to put VPS out of range. That never triggers a "failsafe minimum altitude" preventing the drone from descending again.

The VPS height sensor has no effect on controlling the height of the drone, unless there something is detected within 2ft underneath.

 

[FLDave]

But you suggested that, upon losing VPS height measurement over the water, it triggered some sort of failsafe minimum altitude of 30ft.


Clearly that makes no sense at all, because vertical positioning is controlled by the barometric altimeter, not the VPS sensor.

It makes no difference why it loses the VPS height reference, whether it's over land or water. It will also lose its VPS height reference over land whenever the drone climbs high enough to put VPS out of range. That never triggers a "failsafe minimum altitude" preventing the drone from descending again.

The VPS height sensor has no effect on controlling the height of the drone, unless there something is detected within 2ft underneath.

So you don't have a more plausible explanation as to why it wouldn't descend to less than 30 feet ATL when the infrared distance measurement was lost over the mirror water surface?

 

[Meta4]

The infrared sensor bounces a signal off of whatever is below to measure the drone's current height. But, (anyone who knows better, please correct me if I'm wrong), this sensor only ever comes into play to actively adjust the drone's height if the sensor detects something within (~2ft?) below the drone.

All correct information except that the infrared sensor range is greater.
Infrared Sensor Measurement Range: 0.1-8 m
Hovering Range: 0.5-30

 

[Meta4]

So you don't have a more plausible explanation as to why it wouldn't descend to less than 30 feet ATL when the infrared distance measurement was lost over the mirror water surface?

That's just not a plausible explanation, for a number of reasons that have already been explained.

 

[FLDave]

That's just not a plausible explanation, for a number of reasons that have already been explained.

So, what is your plausible explanation? So far you have not provided one.

 

[Meta4]

So, what is your plausible explanation? So far you have not provided one.

It's complicated ... I don't have time to get into it right now, but will look into it again when I get back later today.
It's better to not put forward an explanation, than to put forward an explanation that goes against the data and the way the drone actually works.

 

[FLDave]

It's complicated ... I don't have time to get into it right now, but will look into it again when I get back later today.
It's better to not put forward an explanation, than to put forward an explanation that goes against the data and the way the drone actually works.

Thanks for the response. If you go back and read my comments on #10 and #13. These were observational and suggestions only. I have no idea how the firmware uses all this information in restricting descent operations. If you have any questions related to those comments I'd be glad to elaborate.

 

[Zbip57]

All correct information except that the infrared sensor range is greater.
Infrared Sensor Measurement Range: 0.1-8 m
Hovering Range: 0.5-30

Agreed. The values from the infrared height sensor, as recorded in the CSV file downloaded from post#4, reach a maximum value of 45.3ft. So it does have an impressive working range.

But what does it actually do with those readings?

As far as I can tell the infrared height sensor never actively does anything useful, unless it senses something at very close range to its underside, within approximately 2ft.

 

[Ranman1022]

Agreed. The values from the infrared height sensor, as recorded in the CSV file downloaded from post#4, reach a maximum value of 45.3ft. So it does have an impressive working range.

But what does it actually do with those readings?

As far as I can tell the infrared height sensor never actively does anything useful, unless it senses something at very close range to its underside, within approximately 2ft.

Good stuff and makes sense except why waste the time with the infrared parts and expense if its only purpose is less than 2ft? Cant the optical handle that like in the Mini2? My Mini2 has always worked flawless. For over 2 years now and im talking over 289 miles of flight logs. ( so you know for reference ) =]

But I thought it switched to VPS at 98ft because the infrared was more accurate than the Barometer. If not then it serves no purpose when the optical can handle accurately handle 2ft as far as i can tell. But on the other hand they might very well have a downward proximity sensor that covers less than 2ft. But ive not read anything about that if they do.

Could it be alignment of the optical sensors 2 patterns similar to how some of the first range finders in WW2 worked. But i doubt it would be that accurate. Those in WW2 were mainly for Battleships and they still had to walk to shots in. But was at great ranges. Up that close it might be very accurate.