Firmware v01.00.0510 EXIF altitude changes

[sar104]

A number of people have noted that the latest firmware has changed the EXIF altitude data, but DJI did not specify exactly what had changed. I guessed that it was what I had been bugging them about – namely that the altitude data fields were mis-labeled, e.g. :

GPS Altitude Ref : Above Sea Level​

Absolute Altitude : +2261.96 – barometric altitude based on a standard atmosphere calculation​

Relative Altitude : +0.00​

GPS Altitude : 2261 m Above Sea Level – the same as the field above – barometric altitude based on a standard atmosphere calculation, not GPS altitude​

I ran a flight test with the previous software (v01.00.04.00), then upgraded to v01.00.0510 and repeated the flight. The altitude data from the DAT files look like this:



The barometric absolute altitude (green) at takeoff is low, by around 120 m (~ 400 ft). The GPS altitude AMSL (blue) is much closer – around 10 meters high at takeoff – but even that error is because only 8 satellites were locked at takeoff. The total went up quickly to 16, and note that the GPS altitude was within a meter or two when it landed. There is no way to verify actual altitude at the high point on the flight, which registered 400 ft on the display, but I would expect that the GPS altitude AMSL was very close at that point too.

This demonstrates, as usual, that the barometric altitude is unreliable, both in absolute terms (around a 400 ft error), but also that the error in relative barometric altitude, which is expected to be linear with relative altitude, can be quite noticeable at 400 ft above takeoff – in this case around 4 meters (13 ft).

Anyway – looking at the EXIF data, with the previous firmware the problem is apparent – both the absolute altitude and the GPS altitude are actually the barometric altitude. With the new firmware, both those fields display the GPS altitude, which is very accurate. The relative altitude appears still to be the relative barometric altitude.

This is a significant improvement, especially in regions with WAAS. Absolute GPS altitude accuracy will be less outside those regions, but will generally still be much more reliably accurate than barometric altitude.

 

[Thomas B]

Good info.
Do you think it will be the gateway to a user reset of the altitude limit on the M2P?

 

[gnirtS]

I'm convinced the Barometer isn't temperature compensated which is why it finishes a flight sometimes significantly different to starting.
This difference is more obvious on a first flight of a sequence where is starting from cold electronics.
Typically I'll see about a 10 to 20ft difference landing vs takeoff (normally negative).
Mavic 1 was the same.

Outside WAAS I'd still trust it more for an absolute than GPS though.

As highlighted by the fact im sat on a beach, next to the ocean so not massively off average mean sea level.
And using 2 mobile and 1 dedicated GPS receivers are showing altitudes of 232, 182 and 176ft respectively.
This is with 20+ satellites available.
Horizontal error is under 10ft on all of them with a completely clear sky view.

 

[sar104]

I'm convinced the Barometer isn't temperature compensated which is why it finishes a flight sometimes significantly different to starting.
This difference is more obvious on a first flight of a sequence where is starting from cold electronics.
Typically I'll see about a 10 to 20ft difference landing vs takeoff (normally negative).
Mavic 1 was the same.

Outside WAAS I'd still trust it more for an absolute than GPS though.

As highlighted by the fact im sat on a beach, next to the ocean so not massively off average mean sea level.
And using 2 mobile and 1 dedicated GPS receivers are showing altitudes of 232, 182 and 176ft respectively.
This is with 20+ satellites available.
Horizontal error is under 10ft on all of them with a completely clear sky view.

I've not seen much evidence of barometric drift that would be attributed to temperature. In the two test flights I just did the starting and finishing barometric altitudes were very close (within 1 m). There's no good reason not to have included temperature compensation.

Your GPS vertical errors are surprisingly large. Are those geoid corrected? With an average distribution of satellites the vertical error should be approximately 1.5 times the horizontal error.

Sea-level atmospheric pressure typically ranges from around 970 mb up to 1040 mb. Converting that to an altitude range gives a variation of around 580 m (2000 ft) – that's why you really don't want to use barometric absolute altitude.

 

[gnirtS]

I've not seen much evidence of barometric drift that would be attributed to temperature. In the two test flights I just did the starting and finishing barometric altitudes were very close (within 1 m). There's no good reason not to have included temperature compensation.

Because its extra money for something that isnt really essential. Its ancient now but on the DJI forum someone actually did measurements on the M1 showing variations with temperature.

I dont think ive ever seen the barometer agree takeoff and landing to within a few feet for the first flight of a sequence. Follow up flights its always closer. Variation is more if i start in a colder environment vs tropics.

Your GPS vertical errors are surprisingly large. Are those geoid corrected? With an average distribution of satellites the vertical error should be approximately 1.5 times the horizontal error.

2 of those are definitely geoid corrected. Im in Asia now but in Europe i get somewhat similar measurements. Even the higher grade hardware on our boats commonly show 60ft altitudes when motoring along the ocean and those units are geoid corrected.

Obviously geoid correction isnt a specific thing - there are countless different models for it and one optimised for a specific region isnt going to be as useful elsewhere.

Ive got no idea if the mavic uses any form of correction and if it does, exactly what it uses.

Sea-level atmospheric pressure typically ranges from around 970 mb up to 1040 mb. Converting that to an altitude range gives a variation of around 580 m (2000 ft) – that's why you really don't want to use barometric absolute altitude.

Working with relatives here is the mavic aviation wise effectively sets a QFE to zero the altitude prior to take off. Generally in the very low altitudes the mavic flies at and in weather conditions suitable for flying (ie not huge forming thunder cells etc) the conversion between relative pressure difference and altitude is pretty good. You should be within a couple of feet or at worst, a few tens of feet.
Admittedly DJI almost certainly use cheap, low grade hardware here so you can expect less accuracy than a proper professional grade instrument. I guess also their logic is the baro height isnt overly important - its accurate "enough" for RTH to be correct but when close to the ground the VPS takes over which is significantly better.
Other than embedding EXIF/SRT i dont think the drone actually uses GPS altitude for anything.

 

[sar104]

Ive got no idea if the mavic uses any form of correction and if it does, exactly what it uses.

Pretty much all modern consumer-grade GNSS units are using WGS84 + EGM96. I'd be surprised if the DJI equipment is using anything else.

Working with relatives here is the mavic aviation wise effectively sets a QFE to zero the altitude prior to take off. Generally in the very low altitudes the mavic flies at and in weather conditions suitable for flying (ie not huge forming thunder cells etc) the conversion between relative pressure difference and altitude is pretty good. You should be within a couple of feet or at worst, a few tens of feet.
Admittedly DJI almost certainly use cheap, low grade hardware here so you can expect less accuracy than a proper professional grade instrument. I guess also their logic is the baro height isnt overly important - its accurate "enough" for RTH to be correct but when close to the ground the VPS takes over which is significantly better.
Other than embedding EXIF/SRT i dont think the drone actually uses GPS altitude for anything.

Yes - but the discussion is not about relative altitude - it's about absolute altitude. Relative barometric altitude is obviously going to be much more accurate than absolute barometric altitude.

 

[paulatkin73]

as this thread on this topic -

New Mavic 2 Pro/Zoom Firmware Released (2019.09.16)

Overview: Date: 2019.09.16Aircraft Firmware: v01.00.0510DJI GO 4 App iOS: v4.3.24DJI GO 4 App Android: v4.3.24 What’s New? Added battery mounting detection. Aircraft is unable to take off if the Intelligent Flight Battery is not mounted correctly. Note that DJI GO 4 v4.3.24 or later is...

forum.dji.com

somebody says in there that DJI fixed smooth transitions for waypoints. i did not test it yet. wonder what else was quietly 'fixed' in there, other than what was posted in the info summary...

 

[GadgetGuy]

I'm convinced the Barometer isn't temperature compensated which is why it finishes a flight sometimes significantly different to starting.
This difference is more obvious on a first flight of a sequence where is starting from cold electronics.
Typically I'll see about a 10 to 20ft difference landing vs takeoff (normally negative).
Mavic 1 was the same.

Outside WAAS I'd still trust it more for an absolute than GPS though.

As highlighted by the fact im sat on a beach, next to the ocean so not massively off average mean sea level.
And using 2 mobile and 1 dedicated GPS receivers are showing altitudes of 232, 182 and 176ft respectively.
This is with 20+ satellites available.
Horizontal error is under 10ft on all of them with a completely clear sky view.

I recall bladestrike telling us that an accurate barometer reading on DJI aircraft would take several minutes to acquire after boot-up, and large discrepancies between takeoff and landing were largely due to not waiting long enough after powering up before launching. If you want a more accurate relative altitudes, let it sit a while longer, even if all other systems are GO.

 

[sar104]

I recall bladestrike telling us that an accurate barometer reading on DJI aircraft would take several minutes to acquire after boot-up, and large discrepancies between takeoff and landing were largely due to not waiting long enough after powering up before launching. If you want a more accurate relative altitudes, let it sit a while longer, even if all other systems are GO.

That may have been true with older aircraft, but the data I posted above clearly shows that's not the case with the M2P.

 

[GadgetGuy]

That may have been true with older aircraft, but the data I posted above clearly shows that's not the case with the M2P.

Thank you for the correction. Glad that has been improved.
When it says Ready to Fly in green, it now clearly is!

 

[ajkm]

Good info.
Do you think it will be the gateway to a user reset of the altitude limit on the M2P?

Can’t see that a user reset of the altitude limit is ever going to be or ever should be possible. There are simply too many cowboys already refusing to follow any regulations - if they ignore VLOS and you then give them the option to fly higher than 400 AGL, do you think they’d stick to the 400ft? We’d be seeing a whole flood of posts on “pushing the altitude record.”

 

[Thomas B]

Can’t see that a user reset of the altitude limit is ever going to be or ever should be possible. There are simply too many cowboys already refusing to follow any regulations - if they ignore VLOS and you then give them the option to fly higher than 400 AGL, do you think they’d stick to the 400ft? We’d be seeing a whole flood of posts on “pushing the altitude record.”

Concerns noted. I’d add that I do have concerns about FAA regs getting more restrictive.

Actually it is possible within current DJI software that comes in the drone. Doesn’t change the drone’s preset limits or modify any programming. I have tested it with my M2P.

 

[ajkm]

In terms of regs, I think we're only seeing the start of it. I suspect the trajectory is going to see the consumer/prosumer drones eventually regulated out of existence as far as individual pilots are concerned and it's only going to be possible to fly as a member of a company or organisation (especially as BVLOS systems are developed). Regs are only as good as the enforcement that goes with them, and there doesn't seem to be much by way of enforcement at all really - though, having said that, the idiot who flew in Toronto was dinged $7,500 eventually (two flights, multiple infractions in both - controlled airspace, night time, over people, advertised event, etc). TC originally though a warning was enough.

At least, as an examiner, it's possible to do some serious education during the process, but anyone being failed by one reviewer can simply book with someone less scrupulous and while the Advanced Ops exam is open book with no required ground school hours, too many are passing without the slightest understanding of the aviation world they're joining (and often a bad attitude as well - "I'll get through this then just do as I please"). Somehow we need to move from "hobby" to "profession".

And yes, totally possible with what we already have.

 

[GadgetGuy]

Concerns noted. I’d add that I do have concerns about FAA regs getting more restrictive.

Actually it is possible within current DJI software that comes in the drone. Doesn’t change the drone’s preset limits or modify any programming. I have tested it with my M2P.

It is also possible with a simple parameter mod on all other DJI aircraft to remove the maximum 500m elevation above launch limit. Almost every DJI drone can already be flown to 1625 feet, which is 1225 feet above 400 feet AGL over flat terrain without modifying anything, by simply accepting the risks in the app of changing the default 120m altitude to 500m. It is the pilot's responsibility to adhere to the 400 foot AGL limit as the terrain changes. Nothing illegal about climbing mountains with your drone from the base, while staying within 400' above the slope under the drone at all times, even though it may be 2000 feet above your launch point!

 

[2020HeinSite]

I realize my response is a bit off topic, but I am wondering where this .0510 firmware update is. I checked my M2P this past weekend and while it did recommend an update it simply did a download and reinstall of 01.00.0400. Which it has now done several times over the past few months.

 

[GadgetGuy]

I realize my response is a bit off topic, but I am wondering where this .0510 firmware update is. I checked my M2P this past weekend and while it did recommend an update it simply did a download and reinstall of 01.00.0400. Which it has now done several times over the past few months.

The newest version of .0400 included the anti-rollback feature, to prevent you from using NLD. That was the only change it made.

 

[HawkEye76]

A number of people have noted that the latest firmware has changed the EXIF altitude data, but DJI did not specify exactly what had changed. I guessed that it was what I had been bugging them about – namely that the altitude data fields were mis-labeled, e.g. :

GPS Altitude Ref : Above Sea Level​

Absolute Altitude : +2261.96 – barometric altitude based on a standard atmosphere calculation​

Relative Altitude : +0.00​

GPS Altitude : 2261 m Above Sea Level – the same as the field above – barometric altitude based on a standard atmosphere calculation, not GPS altitude​

I ran a flight test with the previous software (v01.00.04.00), then upgraded to v01.00.0510 and repeated the flight. The altitude data from the DAT files look like this:

View attachment 82193

The barometric absolute altitude (green) at takeoff is low, by around 120 m (~ 400 ft). The GPS altitude AMSL (blue) is much closer – around 10 meters high at takeoff – but even that error is because only 8 satellites were locked at takeoff. The total went up quickly to 16, and note that the GPS altitude was within a meter or two when it landed. There is no way to verify actual altitude at the high point on the flight, which registered 400 ft on the display, but I would expect that the GPS altitude AMSL was very close at that point too.

This demonstrates, as usual, that the barometric altitude is unreliable, both in absolute terms (around a 400 ft error), but also that the error in relative barometric altitude, which is expected to be linear with relative altitude, can be quite noticeable at 400 ft above takeoff – in this case around 4 meters (13 ft).

Anyway – looking at the EXIF data, with the previous firmware the problem is apparent – both the absolute altitude and the GPS altitude are actually the barometric altitude. With the new firmware, both those fields display the GPS altitude, which is very accurate. The relative altitude appears still to be the relative barometric altitude.

This is a significant improvement, especially in regions with WAAS. Absolute GPS altitude accuracy will be less outside those regions, but will generally still be much more reliably accurate than barometric altitude.

Good info!

 

[star4870]

Can’t see that a user reset of the altitude limit is ever going to be or ever should be possible. There are simply too many cowboys already refusing to follow any regulations - if they ignore VLOS and you then give them the option to fly higher than 400 AGL, do you think they’d stick to the 400ft? We’d be seeing a whole flood of posts on “pushing the altitude record.”

I don't want to start the long discussion and arguing. I'm just disappointed that you guys don't see that this regulations has nothing to do with safety. And because you can't see it, you all deserve it!. I hope in next two years all recreational use of drone will be banned to the point you will be allow to fly only in you own house. But I guess you still will be thinking this is for your safety.

Regards

 

[sar104]

I don't want to start the long discussion and arguing. I'm just disappointed that you guys don't see that this regulations has nothing to do with safety. And because you can't see it, you all deserve it!. I hope in next two years all recreational use of drone will be banned to the point you will be allow to fly only in you own house. But I guess you still will be thinking this is for your safety.

Regards

Can you elaborate? Are you referring to the FAA 400 ft AGL restriction? If so, are you proposing that there should be no limit, or just a higher limit? I'm also curious why you think that all recreational drone use might be banned, let alone why you would hope that happens.

 

[sonof40]

Can’t see that a user reset of the altitude limit is ever going to be or ever should be possible. There are simply too many cowboys already refusing to follow any regulations - if they ignore VLOS and you then give them the option to fly higher than 400 AGL, do you think they’d stick to the 400ft? We’d be seeing a whole flood of posts on “pushing the altitude record.”

I agree 100% but nobody ever addresses the exception. As in my case: scaling a mountainside (stying below 400 feet on the way up)