Hi there everyone!

I'm really quite new to this, all of this. In fact, this is my first post.


I first became interested in this project entirely by chance; I happened upon it while looking for small aircraft that can be used in aerial surveys. I've noticed a few other users have a fair amount of interest in the same topic.


Ultimately I will end up building a drone to capture aerial imagery and use the software I normally use when doing a standard aerial survey (with a much larger Cessna) to produce a DTM. But there are a few things that are still up in the air (...nice pun...) - Life will be so much easier if the IMU could export the inertial measurements at the exact time that the camera takes a photo - is this possible? And to what sort of accuracy?

I know that this is not a prerequisite, but in practice, having your external orientation already computed - in the form of photo ID; x; y; z; omega; phi; kappa - is a real time saver. Rather than manually finding tie-points (because the computer is just hopeless at the task) and having to take ground control.

With the IMU capable of spitting out these readings, a quick survey of an area can be done at low cost with a quick turnaround time. Perfect for disaster relief and the like.


Has anyone had some experience with this? What are the IMU reading like with the ArduPilot?


Many thanks! 


Views: 18518

Reply to This

Replies to This Discussion

Hello Avi !

It is surveying aerophotogrametric software. It is accurate (expensive as well). Accuracy you can get out of it is the size of the ground resolution pixel. There is a lot of manual work that has to be done to get that accuracy but if you need something precise than you have to spend some time on it. So, my finest results are small archaeological sites with the ground resolution of 5mm (1000m2). That accuracy was achieved with the MK octo flying 17m above the ground. The GCP coordinates were taken by the total station with the mm accuracy. Time I have spent on that job was 4 hours of driving to get to the site. One hour of preparations like GCP signalistaion and measuring. 5 minutes of flying. 8 hours were spend to get good results and nice DTM as well as DOP. I did many projects with the 1cm resolution and that's approximate horizontal precision that I have achieved (+/-1cm). Regularly I am doing digital orthophots with the resolution of 5cm covering approx. 25ha in one flight. For those lower resolutions I am using NEX 5 with the 16mm Sony lens. For the higher precision jobs I am using Zeiss Distagon 18mm lens. The reason I am using Sony lens on the higher flights that cover bigger area is the weight of the Zeiss that can be a bit too heavy for the longer flights.

  I have the new Hexa from Jdrones, that doesn't fly for now. Hopefully it will be strong enough to carry the load of two 8A batteries and my new Sony NEX 7 with the Zeiss lens. That would be my final configuration.



Hi Darko

Thanks a lot for the elaborate information.

It seems like you do a good job and get excellent results,

I think however that you might be able to produce all of it much faster and easier using our IPS3.0 SW or "process as a service"

I’d like to give you more details on that, can you give me your mail so I could give more specs?



Hi Avi !

I've peeked at the IPS site. Looks impressive and expensive. Since I have 2 full licences of Photomod, I'll stay with it. The slow process could be just because I wanted to do lot of things manually, like brakelines. The problem with automatic DTM generation is that it doesn't recognize brakelines. It can be very dense but it is never as good as when you make the brakelines manually. I have done few flights with Cesnna too. The camera was Canon EOS 5D II with the Zeiss Distagon 25mm (or 24?) lens. The whole setup was done totally primitive, but working. Camera was hanged on the wing support just beside the door that couldn't be opened totally because of it. The wires were "installed" trough the window or better to say between the window and frame. Now when all those Drones are available I gave up flying around when it is much more comfortable to upload the waypoints and get instant results without paying a cent for the flight. (maybe few cents for battery charge). The only problem is that I cannot advertise that I am doing aerial survey with the drone. Those little "toys" are illegal here. My legal tool is the helium blimp that I used last few years. Watch this panorama to see the results with the blimp.


The Octocopter was a relief and total relaxation after working with the blimp. I don't need a crew of three people, large trailer and a lot of luck with the wind, to finish the job.

I will receive my Sony NEX 7 tomorrow. The lenses are ready and I am hardly waiting for the first test.

My e-mail address is :


That's what I thought as well and I checked.  The problem is definitely on the onboard log side.  If you compare the two attached logs of the same flight from the time of first 3D fix, you'll see that on the ground control side it starts at about 11:28 and the flight ends at about 12:09 which equates to about 41 minutes.  However the onboard log starts at line 1017 with a time of 21335400 ms and ends at line 41727 with a time of 22073900 which is roughly 12 minutes.

If you graph the time of the onboard log, you'll also notice jumps in time.  The biggest one is between lines 32500 and 32502.

I am now flying with Arduplane 2.27 and have noticed the same issue...granted the time lag is not as important but it is still present.  I'm not sure if this is an issue that had been addressed in the passed and solved with 2.28

Any thoughts?


Hi All, Newbie (ish) here... I have a few questions about the photogrammetry proceess, I've had a read through this thread, and a few of the links from it. And i would like to create some digital 3d surface models from aerial photographs. and would just like to get a firm understanding of the process to follow, please could you correct me if i'm wrong,

1. Take a series of photographs facing vertically down from the aircraft. (I believe APM2 has the facility to stabilize a camera gimbal if the plane is not quite level using pitch/roll servos)

2. Use an Orthophoto program to remove distortion in the photos

3. Use a stitching program to stitch the photos together 

4. Insert this large OrthoMosaic into a program that creates a 3d image from it

As a novice, are there any programs i can use to do this manually? Or is the only way to buy a very expensive computer program? or is it best to leave it upto the pros?!

I am in the process of building an aircraft to do this, but it looks a bit complex!

Cheers :)

Hi Richard

You can use P&S for orthophoto, but you have to clear some things first.

1. Yes for the vertical, but stabilization is more DIY at this time.

2. Yes, you have to rectify or orthorectify your images with fancy- easy to use-expensive SW, or use something open source or in lowprice. In order to rectify you need ground control points or base map imagery or use log data with a VERY good sw (leica), also you will need contour lines or dem data for ortho.

3.Use a stitching program only if u wish to have all your images in one. I prefer them separate... 

4. No. SW detects the same (different angle) points from separate images to produce point cloud (if you have 50% overlap the same point will be in 5 photos but from different angle)

Programs: AutoCAD Map, ArcGIS, Erdas imagine, Global mapper, Dronemapper.com (online and very promising), QuantumGIS open source,  and many many other  online servises (search for blog posts) or open source sw (search google) 

building the A/C is more like following rules than tricky or complex. Lift capability for your equipment, stability vs  maneuverability, mapped area etc etc. 

Wish you the best


Hi James, Cheers for the info, It was extremely helpful :-)

1. I am making a small pitch roll plate to attach the canon(CHDK) p and s camera, I have the Canon Ixus 80, and the Canon Ixus 120. I have heard that taking photos every 1.5 - 2 seconds is enough, flying at 200feet AGL, Infnity Focus, Max Resolution etc, no flash etc.

2. It seems that ARC GIS is capable of Orthrectify'ing photographs, It is $100 for a home use licence, so while I am doing this for pleasure, I guess this will do for rectifying the photos.

3. Does the Stitching happen in the final process? 

4. If i use ARC GIS for the orthorecifying, I guess it will also be able to create the DSM/DEM. 

I have already built the airframe for this, the Ursus, capable of flying for over an hour with a decent payload, currently integrating it with APM2, and using Flightgear sim for HIL loop testing, which i'm not having much luck with!

A picture says 1000 words... 

Hi James,

I'm new to all this but by reading this forum I see you guys are doing something similar to what I'm doing. I noticed something in your reply to K.D. Stanhope. Sep 12

"so the only use is aerial survey to clasify species"

Do you know anything about the software to "clasify species" ?

Hello again Richard,

1. Take a look my results using a mod Canon Ixus 100is with CHKD and intervalometer:






2. ArcGIS is a little tricky with lisences... ARCveiw, ARCinfo Are the same thing but with different toolbox.. toolboxes make the difference, for sure you would be able to rectify your photos with the basic tools, but i dont know if you can orthorectify, Point cloud is the most expensive tool both for ARCGIS & Leica geosystems


3. Sticthing is completely  optional, and happens when ever you want, you can use image prossesing SW for that also, like Microsoft IC, you dont need a special photogrametry for that. But its easier to rectify one image (stiched) than all you images separatly.


4. Read 1


Very nice airframe you got there, i can see plenty of room and large wingspan. Forget sim and putting directly all your gear. Make a lot manual RC fligths to ensure that everything works great first, then fly with dummy load, and finaly start sim or APM tests.


Also very nice GC case...

hello Luke,

There are two main clasification methods: Supervised and Unsupervised. Iam using erdas imagine with supervised clasification i.e. I take some known spectrum signatures on ground from known species and then SW searches in my imagery for those signatures using maximum likelihood method. if you search google for remote sensing programs you will find many others..

1. Nice examples of what can be achieved with a relatively cheap little camera, I have got the CHDK installed on my Ixus 80 8MP camera, And ive also go the intervalomter working with good results on  high shutter speed.

2. Hopefully i can get hold of the toolboxes legally for testing and home use, and then licence if I can get good enough results. But i guess you need 'point cloud' in order to create DEM/DSM files.

3. Still don't quite understand the stitching. I guess it wont be possible to stitch non ortho rectified photos. So you need to recitify each one, stitch to one big image, then turn that into 3d.

I've been using the HIL settings to simply learn how to use the PID screen with the Cessna in flightgear (Xplane is in the post!). Which seems to roll upside down on take off in auto mode!

I'll certainly make an hour flight with dummy load with no AP equipment, I guess that should be enough to trim out the aircraft and set correct COG etc. Unfortunately the Ground station was build without a laptop in mind! Id like a small laptop for the field, and for when I get some 3dr radios, any recommendations on the laptop? 

@ Richard and James

Hi Guys, as a Newbie I'm finding this thread extremely interesting, if not confusing, due to the plethora of acronyms. James could you explain further ?

1. Take a series of photographs facing vertically down from the aircraft. (I believe APM2 has the facility to stabilize a camera gimbal if the plane is not quite level using pitch/roll servos)

1. Yes for the vertical, but stabilization is more DIY at this time.

So the APM2 doesn't stabilise in the roll axis ?


Reply to Discussion


DIY Drones Monthly


Season Two of the Trust Time Trial (T3) Contest 
A list of all T3 contests is here. The current round, the Vertical Horizontal one, is here

© 2016   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service