Hi All,

https://app.airdata.com/share/vJcech

Can someone help me figure out what caused this crash? I was flying near a building, but don't believe I struck the building (at least not while the drone was under my control).

Background: Prior to flight, the app required calibration, which seemed to complete successfully. It was bit hard to lock on enough satelites, but it semed stable at 9 when I launched. I had clear line of sight on the drone, besides for when I looked down at controller. It happened fast- my only real recollection is trying to rotate and fly the drone one direction, but it behaving eratically before plumetting.

Mostly I want to determine if it was caused by1. an impact while under my control
2. a mechanical/calibration problem

I have years of flight experience and this has left me a bit shaken. Never had any close calls before.

Thanks!
Noah

Did you have home point set? Number of satellites does not matter, it is their location.

Mobilehomer Posted at 9-17 17:57
Did you have home point set? Number of satellites does not matter, it is their location.

Honestly, nto sure. I normally get the "home point has been updated" message, but can't remeber this flight. It was on GPS mode when launched.

The logs do not show home point set. Tons of warnings. You were in Sport mode when the crash happened.

You had poor GPS for the whole, short flight, due to your choice of location between high-rise buildings which blocked the skyview and satellite reception.
The drone was not able to record a homepoint or record any location data except for some inaccurate tentative data between 4.6 and 6.4 seconds..
You launched, climbed to 100 feet and switched to Sport Mode, flew around a bit and climbed to 217 feet where you crashed the drone into the side of a building at 30.5 seconds and it came tumbling down.

Without GPS your drone had no ability to hold horizontal position and no brakes.
It would continue moving when you took your hands off the joysticks and Obstacle Avoidance couldn't work.
That's fine when you are in a wide, open area with nothing to hit.
But can be a serious problem when you are in potentially windy canyons between buildings and aren't aware that the drone has no GPS.

You explained that as if you were sitting in the drone Labroides. I am in awe of your understanding of these flight statistics and data.

Hi Noah,

Outcome of your flight is the same as Labroides post to you, against a building crash.
But data interpretation is not the same.

Number of GPS satelittes and signal 'health' not the enough to set HomePoint during the whole flight.
Health needs a value of 4 or more (range 0 - 5), in your flight this value for 2, shortly 3.

With low GPS reception the max height is set to different values, see the manual for that.

Your flight started with the stability gained from the vision, not by a GPS lock.

At 3.1 seconds the max height setting set to 5 meter (despite your user setting of 500 meter)
At 4.3 seconds in flight max height set by software to 500 meter. 10 sats + 3 for health.
At 6.1 seconds max height set to 5 meter
At 11.7 seconds back to 500 meter
At 15.5 back to 5
At 21.1 back to 500 meter.

[ 500 meter is your own max height setting, is active/selected by software if flying in a GPS mode ]


At 21.1 back to 1640 feet / 500 meter. (10 sats + 2 health), at this time Sport mode with a GPS lock!
See the proof for this in my chart of your data
- RC pitch input at 22.7 secs, speed increased.
- RC pitch zero input at 24,6 secs, speed decreased ; your Mavic2 from Pitch DOWN to Pitch UP ( only possible if in a GPS lock ! )

for your info : without a GPS or "Vision lock" drone in ATTI mode ( no hover stability ); after relasing RC sticks to zero the drone pitch and roll goes to zero attitude, thus no opposite steering done by software to bring the speed down to zero into a hover.
In ATTI mode speed will eventually slow down in a non-wind situation, with wind it will drift away on the wind.


So your drone nearly stopped on time, but just too close to the building (with some other RC inputs ) it hits the building....

cheers
JJB

Great explanation JJB*. My comments to Labroides also refer to your ability to read and explain the Data provided.  

yogi053 Posted at 9-18 02:21
Great explanation JJB*. My comments to Labroides also refer to your ability to read and explain the Data provided.

Thanks yogi  only my text is correct  ;-)

Thanks all! Much appreciated. I feel better knowing that I contributed to crash, not just a mechanical failure.

So main takeaways (mistakes) are:
1. Ensure home point updated before ascending
2. Launch further from buildings
3. Don’t switch to sport so close to buildings (I normally don’t, but deviated here)

NoahB Posted at 9-18 06:26
Thanks all! Much appreciated. I feel better knowing that I contributed to crash, not just a mechanical failure.

So main takeaways (mistakes) are:

Hi,

for your #1, best to wait for HP to be set before takeoff.
But if (after takeoff with GPS lock) you fly close to high buidlings it is easy to loose some satellites, if you loose enough than your drone into ATTI mode.
So watch carefully the numbers of receiving sats during the flight.

#2 yes, see number #1

#3 yes  high speed flights need more distance to brake.... ;-)

cheers
JJB

JJB* Posted at 9-18 10:33
Hi,

for your #1, best to wait for HP to be set before takeoff.

thx JJB - much appreciated. Would you say not ensuring HP was set was the main cause of crash?

NoahB Posted at 9-18 12:01
thx JJB - much appreciated. Would you say not ensuring HP was set was the main cause of crash?

Would you say not ensuring HP was set was the main cause of crash?
Even if you had a homepoint, that wouldn't have helped you for the rest of the flight where you had poor, unreliable GPS the whole time.
The data that the GPS unit gave did not change at all after 6.4 seconds, despite you using the right stick quite a bit.
The flight controller understood that you had no position or speed information and rated the GPS reliability at 2/5.
Unless it's 4/5 or better, it won't use the GPS information.
Without reliable position and speed data, any attempt to hold position or avoid obstacles cannot work properly.

Flying in a GPS-denied location was always going to give you problems, even if you had started out in the open, recorded a homepoint and then flown in between the tall buildings.

Hello there Noah. I am sorry for the trouble this has caused and thank you for reaching out. Since your DJI Drone crashed before. I would recommend for you to contact our DJI Support Team at https://www.dji.com/support?site=brandsite&from=nav for further assistance. We will do our best to help you and give out the best resolution for the said issue. DJI will perform a data analysis service for products under warranty to determine the cause of the accident and provide the proper resolution. If it is evaluated as a non-pilot error, the warranty service will be provided. Again, I am sorry and thank you.

NoahB Posted at 9-18 12:01
thx JJB - much appreciated. Would you say not ensuring HP was set was the main cause of crash?

Hi Noah,

The response of Labroides is incorrect, sorry to say so.

GPS reception to get a GPS lock needs not 4 or 5 but minimum 3 for GPS 'health'.
HP to be set needs 4 or 5.

GPS 'health' is found in the CSV of a flightlog, fieldname =  OSD.gpsLevel (Phantomhelp naming) . Range 0 to 5, this values represents a DJI logarithm of the signal use of satellites.(i.e. receiving 12 satelittes close to each other in the sky is not good enough to get 3 or more....8 receiving sats widely spread in the sky does give 3 or more)
0-2 = NO GPS lock
3 or more = GPS lock
4 or more = HP is set.

As showed in my chart in my previous post, your drone had a GPS lock in part of the Sport mode flight ; that is why it did brake after releasing forward stick.

Another proof is the max height setting wich changed few times from 5 (GPS weak) to your own max height setting of 500 meter (GPS lock).
5 meter is set if GPS reception is weak (GPS data indeed unreliable) but set back to your own max height limit of 500 meter means GPS reception is good enough for a GPS lock.
Not for HP to be set, as drone need 4 or 5.

If a DJI drone cannot use GPS it will automatically enters ATTI mode, well   not in your flight.

About your question ; yes, if you had wait for HP to be set than ofcourse your drone had a good GPS lock at start of the flight. Wich, as i mention earlier, say nothing about the GPS reception (loosing sats) flying close to high buildings etc.

cheers
JJB

DJI Stephen Posted at 9-18 20:58
Hello there Noah. I am sorry for the trouble this has caused and thank you for reaching out. Since your DJI Drone crashed before. I would recommend for you to contact our DJI Support Team at https://www.dji.com/support?site=brandsite&from=nav for further assistance. We will do our best to help you and give out the best resolution for the said issue. DJI will perform a data analysis service for products under warranty to determine the cause of the accident and provide the proper resolution. If it is evaluated as a non-pilot error, the warranty service will be provided. Again, I am sorry and thank you.

Hi Stephen -- This thread has been very helpful, but I'll contact DJI support through that link now. I'm very interested in the analysis and learning all I can to reduce the chance of this happening again -- thanks!

JJB* Posted at 9-18 23:33
Hi Noah,

The response of Labroides is incorrect, sorry to say so.

The response of Labroides is incorrect, sorry to say so.
Disagree all you like, but what the actual data shows is .....
The position data from the GPS unit did not change at all after 6.4 seconds, despite you using the right stick quite a bit.
The flight controller understood that there was no accurate position or speed information and rated the GPS reliability at 2/5.
If the position data is inaccurate, it's not possible for the flight controller to properly hold position or avoid obstacles.

DJI Stephen Posted at 9-18 20:58
Hello there Noah. I am sorry for the trouble this has caused and thank you for reaching out. Since your DJI Drone crashed before. I would recommend for you to contact our DJI Support Team at https://www.dji.com/support?site=brandsite&from=nav for further assistance. We will do our best to help you and give out the best resolution for the said issue. DJI will perform a data analysis service for products under warranty to determine the cause of the accident and provide the proper resolution. If it is evaluated as a non-pilot error, the warranty service will be provided. Again, I am sorry and thank you.

Since the drone involved was a Mavic 2, it won't have been covered by warranty and DJI is unlikely to be much help.

Labroides Posted at 9-19 02:42
The response of Labroides is incorrect, sorry to say so.
Disagree all you like, but what the actual data shows is .....
The position data from the GPS unit did not change at all after 6.4 seconds, despite you using the right stick quite a bit.

Hi Labroides,

Well...
Just explain why his mavic2 did brake after releasing forward stick......see my post #7 top chart with braking data.
Only possibility to brake if drone is in a GPS mode.
Let alone the other facts i wrote.plus see the  OSD.isGPSUsed in the CSV  : true and false  , true for all the same lat lon gps data.

Flying with low number of sats between building ect, then GPS lat lon data is written with spurious values, if you calc distance between data points than you will see variation in distances out of the normal. This is what you mention as "inaccurate"
For some reason is this log no data is written for GPS data, all positions are the same for many records. Don`t know why data isn`t written in the log.
But Phantomhelp does copy paste previous data into next records if there no value in the original txt file.
In similar files, low # sats and/or low in 'health' GPS data is written.



cheers
JJB

JJB* Posted at 9-19 03:29
Hi Labroides,

Well...

Just explain why his mavic2 did brake after releasing forward stick......see my post #7 top chart with braking data.
Only possibility to brake if drone is in a GPS mode.
How this all works is not specified anywhere so we try to work out what's going on but what information we have is confusing.
The drone also has the accelerometers to sense motion, so it may use those to some extent.
But when inaccurate GPS data is used in the fusion, any braking, obstacle avoidance etc will also be inaccurate.

Labroides Posted at 9-19 04:17
Just explain why his mavic2 did brake after releasing forward stick......see my post #7 top chart with braking data.
Only possibility to brake if drone is in a GPS mode.
How this all works is not specified anywhere so we try to work out what's going on but what information we have is confusing.

Oke....  but accelerometers do work also in ATTI mode....no braking in ATTI after releasing sticks, drone pittch and roll to flat values.
In this log typical braking ; from drone pitch down to up to normal.
In the log also True for GPS use, so for me only confusing why the position GPS data is not written in the log.  With low GPS and low quality LatLon is normally written.
Same for ATTI mode when GPS data is not used as choice of software, but data is available and written in the log.

GPS data was not inaccurate in this log, simply not written. So mayby inaccurate but not 'enough' inaccurate to disable the use of GPS data for the whole flight.
cheers
JJB

JJB* Posted at 9-19 05:37
Oke....  but accelerometers do work also in ATTI mode....no braking in ATTI after releasing sticks, drone pittch and roll to flat values.
In this log typical braking ; from drone pitch down to up to normal.
In the log also True for GPS use, so for me only confusing why the position GPS data is not written in the log.  With low GPS and low quality LatLon is normally written.

You've lost me.
The satellite numbers were low.
The amount of sky was heavily reduced by buildings, giving a poor spread of the satellites that were visible.
The flight controller calculated the GPS reliability 2/5 and too low to record a home point.
The app didn't record GPS data or it recorded the same point despite drone movement.
But you suggest everything was fine onboard and the drone was able to fly normally?

NoahB Posted at 9-18 23:42
Hi Stephen -- This thread has been very helpful, but I'll contact DJI support through that link now. I'm very interested in the analysis and learning all I can to reduce the chance of this happening again -- thanks!

Hi there Noah. Thank you for the reply and for this given update. If you have any other inquiries or any concerns with DJI. Please feel free to post it here at DJI Forum. We are all here to help and assist you.

Labroides Posted at 9-19 13:03
You've lost me.
The satellite numbers were low.
The amount of sky was heavily reduced by buildings, giving a poor spread of the satellites that were visible.

You've lost me.Oops, sorry   will bring you back

The satellite numbers were low.
True, but not to low to get a GPS lock,  7, 8 or more can give a GPS lock, even with a low 0-5 number.
DJI software knows this too, as with bad GPS max height is limited (with or without vision in use), so the change form 5 to user max height 500 meter shows that some part of the flight GPS was not that bad.


The amount of sky was heavily reduced by buildings, giving a poor spread of the satellites that were visible.
In theory yes, high building obscure sats, but guess we both don`t know wich sats in the sky and their positions were received.

As said, switching to back from 5 to 500 user meter max height does say it all about GPS reception.

The flight controller calculated the GPS reliability 2/5 and too low to record a home point.
True, need 4 or 5 to get HP to be set, but without HP being sat DJI drones can have a GPS lock.

The app didn't record GPS data or it recorded the same point despite drone movement.
Indeed, but with bad GPS reception normally data is written in the log, only the LatLon positions doesn`t show the actual positions.
In this log GPS lat lon is for many records the same , a copy paste action of phantomhelp if data is not there.
Question is  why not ?? As in the log you can see if GPS is used or not. GPS was uses part of the flight.

But you suggest everything was fine onboard and the drone was able to fly normally?

Nah, i do not suggest that. But data shows that at25s drone was flying in Sport mode with nose down and forward stick at 100% shortly.
GPS sats 8 and 9. After releasing stick brake action, nose up, reduce speed, nose down, nose normal to hover.
Only possible if there is a GPS lock !
Acc meters does not do this, same for attitude measurement by the system.

Plus if GPS is unreliable for the sofware, it will switch to ATTI mode. It did not in this flight.

But agree, not a clever flight to do so...


cheers
JJB

JJB* Posted at 9-19 23:31
You've lost me.Oops, sorry   will bring you back

The satellite numbers were low.

The satellite numbers were low.
True, but not to low to get a GPS lock,  7, 8 or more can give a GPS lock, even with a low 0-5 number.
You know that number of sats alone is not what's required for good reception.
The drone needs good numbers and a good spread.
There were a number of tall buildings around and only a small part of the sky available = very poor spread.

But data shows that at25s drone was flying in Sport mode with nose down and forward stick at 100% shortly.
In Atti, the drone will do that just as it does in Sport Mode too.

The height limits are tricky, but so are other things.For mine, it looks like a false indication of GPS mode, when actually in atti mode.
That might be responsible for the height limits.

Labroides Posted at 9-20 00:12
The satellite numbers were low.
True, but not to low to get a GPS lock,  7, 8 or more can give a GPS lock, even with a low 0-5 number.
You know that number of sats alone is not what's required for good reception.

Hi Labroides,

Yes i know that (about # sats and what is needed for good reception)
But if some part of the sky is blocked by buidlings doesn`t automatically mean that the remaining receiving sats are not, in that part of the sky, spread out enough to give a gps lock.
So your conclusion is bit too fixed, but in general you are correct.

Atti mode and Sport mode (in GPS lock) acts totally different after releasing a RC input.

Assuming no wind
In Sport or Normal mode (gps lock) ; forward flying = stick forward = craft pitch down | stick to zero = craft pitch up to brake > speed to zero, at zero = craft flat pitch
In ATTI mode (no gps lock) ; forward flying = stick forward = craft pitch down | stick to zero = craft pitch to flat pitch > speed slowly to zero after some time

DJI software is clever enough to switch to ATTI if GPS data in unavailabe / not trusted, all other values in the CSV (gps use etc) doesn`t indicate that drone was in ATTI mode without writing this in the log.

Will check if ATTI mode do change the max height setting, afaik / remember it does not change that

So we agree to disagree  ;-)

cheers
JJB