3cm absolute accuracy without any ground control points using Agisoft

[John_4dm]

Just wanted to share an exciting technology that enables any drone to produce 3cm absolute mapping accuracy without GCPs. This video shows check points residuals' using Agisoft Photoscan Professional:
 https://www.youtube.com/watch?v=nT4LRm4FAxE

A white paper available in here: https://www.gim-international.com/content/news/precise-uav-camera-positioning-for-3d-mapping

[stihl]

It's worth mentioning how this system works:

PPK stands for post processing kinematic and works by using one or more Base-stations that uses correction signals to refine the GPS coordinates of a UAV. A big plus is if your GPS-receiver retrieves both L1 and L2 frequencies.
Same as RTK-GPS found in surveying GNSS systems but isn't real time as the name suggests.

A flaw that PPK has, which I found from experience, is that it doesn't always work flawlessly and some camera stations end up less accurate. I always end up using about 4-5 GCPs just to be safe side.
On top of that you still need to measure Check points in the field.

It will be interesting to see how this technique will develop in the upcoming years. 

[John_4dm]

All valid and good points. Some more technical details which would clarify the achievable accuracy:

• Quality and sensitivity of the GNSS receiver used in building a PPK system plays the most important role in getting a robust and clean solution for all camera stations. The one that I was referring has been built based on latest generation of triple frequency Novatel GNSS receivers with highest performance in airborne application.(Export restricted though).
  
  Second important point is precise time synchronisation between camera exposure at the very middle of the exposure time and the GPS time. There are lots of factors and internal electronic delays involved in this, mentioned PPK system is well engineered to take care of all of these factors in a smart way.  There is a one time calibration procedure, which effectively models all of these factors. For the calibration flight a good network of GCPs is required, this procedure is same procedure as calibrating a million dollar metric aerial camera and its direct geo-referencing system.
  
  Once the PPK system calibrated, correction parameters are applied autonomously to any other flights which then provides very reliable accuracy without using any GCPs.
  
  Regarding check points requirement, I'd say check points are just for cross checking and proving the claimed accuracy. They are not used in photogrammetry at all for improving the results, just used for monitoring the errors and totally optional.
  
  From my 20 years of photogrammetry experience with drones and manned aircraft, a good surveyor must always cross check the results irrespective of what technology is used. I always use one or two GCPs to to make sure that camera focal length variation, for example due to temperature variation, and operator errors are taken care in photogrammetry and always get better than 3cm absolute accuracy. 
  

A good thing is all of these issues are well considered in the mentioned PPK system and in its post processing software , so just by following the provided calibration procedure, users don't need to worry about all of these complexities. (same as if you don't calibrate your autopilot you wouldn't be able to get desirable flight pattern).

 



 

[Diego]

Totally agree with user stihl, we must make this clear, and it is always important to at least 4 to 5 points for quality control and safety, can not speak so simply considering that there are many inexperienced people in this forum and can be with a wrong idea, It is important to use one or more base stations for signal correction and to improve the GPS coordinates.

[John_4dm]

That's correct. PPK system requires a base station for post processing. Similar to RTK (real time Kinematic) but doesn't need to receive the base station correction via radio link while flying. See some more detailed information in the product page here: http://geomatics.com.au/

[frank.stremke]

sound very good
i was using something similar home build for kite photography but since i never got the time calibration right i deserted the project.
so anyone know that the price tag might be on the UAV version minus the ground station since i can just use my leica GNSS for that
frank

[Sirius]

Frank - I paid 100k (new Zealand Dollars) for my RTK enabled Sirius Pro UAV a couple of years ago.  They may be cheaper now.  That includes software for planning etc and a Photoscan plug in to make it all work.
As mentioned previously - still need to put in a few check marks to make sure its working and to make sure you didn't do something dumb like measure the base height wrong.
The bolt on RTK / PKK soloution mentioned here would be a lot cheaper I assume.  I might look at pricing one for my non RTK multirotor.

[peha]

Is this technology also applicable for  the ebee from sensefly? Or does it need further hardware?

[John_4dm]

Yes it is compatible with ebee, Sirius, Falcon and any other airframe, according to http://www.riseabove.com.au/ppk-gnss-system-for-multirotor-and-fixed-wing-uav