REQUEST TO DJI: Photogrammetric GPS and IMU data

[RotorTalk]

In order to undertake aerial photogrammetric data processing (orthorectification/image mosaicing) its necessary (preferred) for the 3D position of the camera, its attitude and the time the picture is taken to be recorded as part of the data capture mission.

Companies such as Sensefly, Pteryx, Pix4d etc incorporate this ability in their flight/camera/GPS AP systems.

If DJI is really serious about providing AP/photogrammetry capabilities then at the very least one would need a data log recording the above parameters using the Wookong-M (Ace-one etc) controllers. Another way would be to write this data into the camera's file as part of each photo's EXIF/IPTC metadata file.

As an interim fix can one get a dump file out of DJI 'View'? Then one could get by by syncing the cameras time to GPS time.

 

[ChrisViperM]

DJI acts like a stubborn old wife.....whatever she says is right, and whatever you are saying or wanting is bullsh... unless approved by the old lady.....

....data log recording is a feature what users are asking for now for a long time (and not difficult to implement) , but DJI thinks we don't need that....full stop. Sad, but a proven fact.


Chris

 

[srbell]

Again, my $179 APM 2.5's and $89 40km RDF900 modems are kicking my MKm/DataView in the butt. I think they better look at the competition... And what's with the $1000 retractable landing skids? Wow, that's a bit steep!

 

[matthew.wright5]

It will come eventually - DJI is a lot like Apple computers - they have the tech now, but release it in increments over a prolonged period to keep us interested and themselves in the news and at the top of the game.

They have the option to charge for the capability too!

That said, I would like to do photogammetry in the future, so DJI better prove me right!

 

[Dewster]

Photogrammetric data processing?? I just want to put a camera up in the air and take pictures. I'll leave the innovation to MIT students. I think DJI found it's customer base by providing a relatively simple flight system that allows most to fly and take aerial photos. Despite professional labeling and incredible flight performance, I don't think this gear is really intended for scientific research. It's a simple flight platform. You'll need to attach your specialized gear if you want scientific measurements etc.

I do think the $1000 dollar price tag on retractable landing gear is steep. It's just unnecessary. The retracts on the landing gear are pretty cool and something that I don't think would be hard to do with a little ingenuity.

 

[matthew.wright5]

The actual photogrammetric processing is done by other programs. The least DJI can do is provide a exportable doc that contains a list of GPS coordinates that corresponds to the location where each picture was taken.

These programs use that data for useful mapping applications and analysis. Without it, all you have is a stitched landscape picture of the ground.

 

[Muunk]

My Sony A77 camera has a built in gps

 

[srbell]

The A77 is a nice camera, too bad it's double the weight of a NEX. Maybe the next NEX will include a GPS...

 

[ChopperBoy]

How to configure the "F-Channel" and "GP servo action config" for using the "Photogrammetry Tool"? Anyone tried with S800 + Z15 + Nex7 and got successful automatic photo taken at predefined photo point?
We'd tried to define a flight mission using the photogrammetry tool, inputting H/V overlapping %, flight speed ...., there were estimated total flight time and length, however the estimated no. of photos is always zero. I assume that once the flight mission is upload to S800, the Main Controller will send signal to Z15 to trigger the infra-red shutter for photo taking, no extra cable is required, am I right?

 

[Digitech]

there will be a log function soon comming with the iosd
it will store all parameters.
if you by chance destroyed a iosd you can see there is a memory card inside.
or maybe its for DJI to see what you are all up to these days

 

[uavservices]

That logged data will be very inaccurate.

Yes you can solve the camera position faster but that is about it. Final DEM model creation MUST make use of proper and accurate ground control points. So if you want an accurate model that can be proven to your client you better go buy 20-30,000$ in survery equipment and learn how to use it... as me how I know

You can make a model with camera position data but nobody in their right mind will pay you for it if you can't actually prove the data and show them the residual error.

It is a "nice to have" but the last thing I want is this thing hanging up becuase it's logging a GPS position that is at best within 3-5m accuaracy. Typically my GCP's are surveyed to within less than 1.0cm accuracy (or better) and the more accurate your GCPs are the more accurate your model is...

I don't typically log data onboard any of my aircraft

If you are looking at selling the data you are aquiring in an accurate manner this is the wrong way to go about doing it. Yes it's cheaper than buying survey equipment and learning how to use it... but proven data is worth money, unproven data is just "cool"

Steve

 

[uavservices]

if you really want the gps position of each photo you can buy a gps logger, sync it with your cameras time, then you can update the exif file of your photos by syncing them with the logged gps data. This solution does not put any more "strain" on the WKM brain and can also be applied to other systems of your liking...

I have this http://blip.com.au/item.aspx?itemid=62 and it works well.... I have not used it in ages but as I recall it can log fairly quickly and it may even come with a syncing utility. As you can see accuaracy is 5.0m, even if it was WAAS enabled It may get down to 3.0m


Steve

 

[Dewster]

Lol.

 

[kerbob]

I'm new to the Wookong-M Waypoint game but have had a $189 APM 2 for many months. I just assumed the Wookong-M could do the same thing.. like export kml files and show me the height, speed, orientation, and voltage along any point along the way. This really does suck.. I really learned a lot from those logs.
The Wookong-M is still ahead of the APM 2 on stability.. but the gap is closing.

 

[srbell]

I'm new to the Wookong-M Waypoint game but have had a $189 APM 2 for many months. I just assumed the Wookong-M could do the same thing.. like export kml files and show me the height, speed, orientation, and voltage along any point along the way. This really does suck.. I really learned a lot from those logs.
The Wookong-M is still ahead of the APM 2 on stability.. but the gap is closing.

I'd have to say stability is about on par now. I also have a APM 2.0 and a few 2.5 boards. Combine that with the RDF 900 radio model with 40Km range and you have a killer combination. I'm going to be doing more comaprisons with my wookong APM's in the next couple of weeks. My S800 may have a AMP soon lol.

 

[uavservices]

very interesting guys, I have up on Ardu anything years ago because it was very unreliable.... interesting to hear it has come a long way.... One day I may have to try it again.

Thanks for the info and opinions on the APM

Steve

 

[remyd]

DJI do log the data using their Aceone products. Unfortunately they don't allow the .dat files to be read by users rather will view the flight logs when there are problems to help sort them. Apparently the Aceone single waypoint logs are fairly basic while the multiwaypoint uses a completely different box and has more logging going on. I do agree with other that this IMU data should be readable and it certainly wouldn't be difficult to write a small program to allow access to some of the data and hide what DJI don't want us to see. My understanding is that the wookong IMU data isn't accessible and while the iOSD might allow simple things like GPS location and altitude the true IMU that you are after is unlikely to come through the iOSD.

Steve - is right a simple external GPS connected to the camera can be utilised by cameras and then the some of the programs such Pix4d can read the exif header but i disagree that the data from these programs are not useful. Yes they may need some additional accurate 1-2cm GCP's to help rectify the image but some of the models being produced from these techniques and from others based around structure from motion are on par with DEM's produced by lower grade Lidar and particularly useful in modelling canopies and other structures.

That said i suspect the IMU data flowing from all DJI products will contain fairly high errors, due to low grade MEMS sensors and because normal GPS is used rather than RTK or other 1-2cm highly accurate data be fairly useless for real reconstruction using traditional IMU based techniques. Also don't forget there are more complexities if you are using a gimbal rather than fixing the camera directly to the frame where the IMU is measuring.

All that said i wouldn't mind seeing and accessing the IMU data to get an idea of stability of different setups rather than looking at them and thinking they are more stable than others.

 

[pls4air]

is anyone familiar with the GS photometry servo setting to trigger the camera? we are using sony nex 7. i can take a picture by clicking trigger and reference photo but it will not take photos automatically in flight. i can click reference photo while we are in flight to take pics but really need it to take pics at the planned times.

 

[srbell]

Set your servo for 900, -900 and a short delay. But you'll need to use the photogrammetry function, not the scan to trigger the camera. just before you upload you should see the action in meters. Unfortunately, if you're also using a Zen with that Nex7 (like we are), you'll get usually two exposure for each action point. A pain. Almost as much as the fact the iOSD won't record time stamps for syncing the photo's gps data! Arrrrg! Still using a data logger.

is anyone familiar with the GS photometry servo setting to trigger the camera? we are using sony nex 7. i can take a picture by clicking trigger and reference photo but it will not take photos automatically in flight. i can click reference photo while we are in flight to take pics but really need it to take pics at the planned times.

 

[leofuica]

Originally Posted by pls4air
is anyone familiar with the GS photometry servo setting to trigger the camera? we are using sony nex 7. i can take a picture by clicking trigger and reference photo but it will not take photos automatically in flight. i can click reference photo while we are in flight to take pics but really need it to take pics at the planned times.


Hi,
first you need fill all data for Photogrammetric , sensor size, overlap H and W, lens, altitude, etc.
Next you can scan area and preview, in this point you can see a numbers of photos will be take (if you can see this the camera will be takea pics in flight).
If you agree push export and edit again altitude, h speed, v speed, turn mode, etc.
Save and upload a mission and GO...

I hope my explanation helps you....