Auto-landing when descending below take-off altitude.

[Meta4]

Given the sensors are ultrasonic is there a possibility of the falls sound bouncing off the valley and disrupting the sensor?

That's unlikely, if this was a problem with the system, others would have noted it by now.
People fly out over cliff edges all the time and many have flown from the same locations you have.
I know because I used to live quite close to there myself.
You had an issue with your VPS sensors reading something that wasn't there.
Perhaps it was an intermittent issue (?) but it's not related to the flight location.

How did you get the drone back?
The flight data ends with it still out and doing another forced landing.

 

[ManBearPig1989]

That's unlikely, if this was a problem with the system, others would have noted it by now.
People fly out over cliff edges all the time and many have flown from the same locations you have.
I know because I used to live quite close to there myself.
You had an issue with your VPS sensors reading something that wasn't there.
Perhaps it was an intermittent issue (?) but it's not related to the flight location.

How did you get the drone back?
The flight data ends with it still out and doing another forced landing.

I landed at a lookout where I was standing - away from the takeoff position. If it's intermittent it sure is consistent with the two waterfalls. Maybe it's a fault that only presents itself in a specific circumstance.

 

[Meta4]

I landed at a lookout where I was standing - away from the takeoff position. If it's intermittent it sure is consistent with the two waterfalls. Maybe it's a fault that only presents itself in a specific circumstance.

There's nothing about the locations that would cause the issue.

 

[ManBearPig1989]

There's nothing about the locations that would cause the issue.

Then I am at a loss as to why out of the hundred so times I've flown, the only three times this has occurred has all been at the exact same location, all at different dates. Hence why all my ideas have been based around either water vapour, sound or the valley causing the disruption. Maybe its a calibration issue that only presents itself in this scenario? Again I'm just throwing out ideas at this point as I'm utterly lost. Did you live near Mapleton? If you ever do find yourself flying at Mapleton falls off of the lookout let me know if you experience this.

 

[Meta4]

Then I am at a loss as to why out of the hundred so times I've flown, the only three times this has occurred has all been at the exact same location, all at different dates.

And of those hundred flights, when were the three that you noticed the issue?
All fairly recent?

You had an error message show twice in this flight at 2:57 and 8:12.7. It showed:
Downward sensor error. Aircraft unstable at low altitude. Please fly with caution.

The forced landing only showed when you held the left stick down close to the bottom for a second or more.
You might not have done this in other recent flights or if you did it was when you were bringing the drone down to land and not noticed it.
Your flight data from May 24th did not show any false VPS readings.
If it was my drone I'd be very concerned about the issue and want to check it out properly.

Hence why all my ideas have been based around either water vapour, sound or the valley causing the disruption. Maybe its a calibration issue that only presents itself in this scenario?

Sounds and the valley would not have been a factor.
How close to any splashing water was the drone?

 

[sar104]

The Kondalila flight was not the first time this has happened. Every single Mapleton falls flight this has occurred. The home flight was after all the Mapleton falls flights which gave the exact same issue. Every single flight away from those two problem locations function absolutely fine. I actually had a flight on my tablet when I got home, I'll upload that when I get home. The entire reason I uploaded the other home flight was to show that the sensor issue only occurs at the falls. Given the sensors are ultrasonic is there a possibility of the falls sound bouncing off the valley and disrupting the sensor?

For the flight that you posted, the problem is perfectly clear. The bottom sensor measured zero altitude for the entire flight, as shown above in post #9. As a result, every time you applied full down throttle, the aircraft switched to landing mode. It has nothing to do with National Parks or restricted flight zones (which National Parks are not). The faulty sensor readings are unlikely to be due to the terrain. If you don't have something installed on the aircraft that is confusing the sensor then you have a faulty sensor.

 

[ManBearPig1989]

And of those hundred flights, when were the three that you noticed the issue?
All fairly recent?

You had an error message show twice in this flight at 2:57 and 8:12.7. It showed:
Downward sensor error. Aircraft unstable at low altitude. Please fly with caution.

The forced landing only showed when you held the left stick down close to the bottom for a second or more.
You might not have done this in other recent flights or if you did it was when you were bringing the drone down to land and not noticed it.
Your flight data from May 24th did not show any false VPS readings.
If it was my drone I'd be very concerned about the issue and want to check it out properly.

Sounds and the valley would not have been a factor.
How close to any splashing water was the drone?

I have definitely held the throttle stick downward for much longer than a second in numerous other flights - many of which were to come down from a higher altitude. I suspect it really is a sensor issue and I'll start up a case with DJI. I have re-calibrated multiple times both in app and through the assistant program. I just took the drone for a flight from home and flew around both ascending and descending just fine. However, due to my house being on a hill I was able to take the drone down near the river where I was able to lower the drone below the take-off altitude. As soon as I did this the VPS started detecting the ground at 0.2m etc. despite still being 20 or so meters above the ground. This forced it into landing mode and descended another 15m or so before the VPS accurately detected the ground at 6m. As soon as I flew back up above take off position the VPS issue (detecting ground at 0.x meters) stuck seemed to remain for the rest of the flight. Why on earth this issue kicks in when descending below take-off altitude now is beyond me.

 

[Meta4]

As soon as I did this the VPS started detecting the ground at 0.2m etc. despite still being 20 or so meters above the ground. This forced it into landing mode and descended another 15m or so before the VPS accurately detected the ground at 6m. As soon as I flew back up above take off position the VPS issue (detecting ground at 0.x meters) stuck seemed to remain for the rest of the flight. Why on earth this issue kicks in when descending below take-off altitude now is beyond me.

Together with the error messages that sounds like a faulty sensor.
You could always just disable the VPS sensors in the app settings, since there's not much you really need them for anyway.

 

[ManBearPig1989]

Together with the error messages that sounds like a faulty sensor.
You could always just disable the VPS sensors in the app settings, since there's not much you really need them for anyway.

Don't they engage the landing? or is that not the case? I'd rather not throttle straight down into the ground - but given the alternative is sending my drone off to get repaired a fair way away I'd rather just have it functioning respectably.

 

[msinger]

Don't they engage the landing?

Yes. If you disable the Landing Protection setting in DJI GO, you'll have full control of the Mavic until it makes contact with the ground.

 

[DanMan32]

M2 does not use ultrasonic. It is IR.
I'm assuming OP's quad is an M2 since he's posted in an M2 forum.

Could be sensor calibration issue. It does have stereo optical as well.

 

[ManBearPig1989]

M2 does not use ultrasonic. It is IR.
I'm assuming OP's quad is an M2 since he's posted in an M2 forum.

Could be sensor calibration issue. It does have stereo optical as well.

Yeah it's an M2. I thought maybe it was a software issue related to NLD, given I've never had this issue before and it only manifests itself arbitrarily based on take-off altitude. I've tried re-flashing the drones firmware and going through the calibration process in assistant 2 with no such luck unfortunately. But strangely when checking the HD graph it shows I'm still in FCC so I wondered if it somehow hadn't flashed properly. Either way, I will check again tomorrow to see if I can get it thoroughly reset.

 

[sar104]

Yeah it's an M2. I thought maybe it was a software issue related to NLD, given I've never had this issue before and it only manifests itself arbitrarily based on take-off altitude. I've tried re-flashing the drones firmware and going through the calibration process in assistant 2 with no such luck unfortunately. But strangely when checking the HD graph it shows I'm still in FCC so I wondered if it somehow hadn't flashed properly. Either way, I will check again tomorrow to see if I can get it thoroughly reset.

Well if you are running hacked firmware then all bets are off. It would have been helpful to have mentioned that up front.

 

[ManBearPig1989]

Well if you are running hacked firmware then all bets are off. It would have been helpful to have mentioned that up front.

I didnt mention it because this issue has been occuring before NLD at Mapleton too. I have since reflashed the firmware with no changes.

 

[JAW]

Could it be the dreaded dOn't_haCk-mE bug?

 

[Montana John]

Correction ... It used to fly normally at home.
You need to do some testing now, after the problem flight rather than relying on a flight done a week ago.
The bottom facing sensors are ultrasonic and as shown were giving a false reading for the Kondalilla flight.
That is indisputable as it's clearly shown in the data.
The important question you need to find an answer for is: Are the VPS sensors still giving false data?

The Kondalila flight was not the first time this has happened. Every single Mapleton falls flight this has occurred. The home flight was after all the Mapleton falls flights which gave the exact same issue. Every single flight away from those two problem locations function absolutely fine. I actually had a flight on my tablet when I got home, I'll upload that when I get home. The entire reason I uploaded the other home flight was to show that the sensor issue only occurs at the falls. Given the sensors are ultrasonic is there a possibility of the falls sound bouncing off the valley and disrupting the sensor?
[/QUOTE]
Water vapor? Or vapour?

 

[kilomikebravo]

I believe that one of the experts solved the mystery of the bottom sensors in a previous post...they are infrared, not ultrasonic.

 

[ManBearPig1989]

Could it be the dreaded dOn't_haCk-mE bug?

Nope. As mentioned earlier - the issue was occuring prior to trying NLD. I have reflashed the firmware with no luck.

 

[FN Bushcraft]

I am wondering if the downward vision sensors may be confused by mist in the air from the falls. Perhaps mist droplets with sun shining on them causing them to appear to be more substantial than they are.

 

[sar104]

I am wondering if the downward vision sensors may be confused by mist in the air from the falls. Perhaps mist droplets with sun shining on them causing them to appear to be more substantial than they are.

Water droplets and mist are generally quite transparent in the infrared spectrum due to the wavelength being longer relative to the particle size, but I guess that cannot be ruled out without testing. Note that the VPS height for the flight posted was pretty much exactly zero for the entire duration - to me that suggests a sensor fault.