Dear RTK -users!
For the past 3 years,  Phantom P4P drones have been used in our fieldresearch, gathering data on gray-whale mother calf pairs and photo ID on gray whale individuals. As photogrammetry may provide data on length / width and possibilities to calculate volume, we assume that we could be able to use Phantom RTK drones to be able to assess body condition of these whales.

For those of you who have been using this type of drone within photogrammetry (- i.e. research programs) : how exact is the RTK in practice in its vertical and horizontal measurements? Is it indeed the 1,5 - 2 cm as given in the product description?

I am aware of modified DJI equipment with third-party components as a work-around to obtain better and more reliable height/elevation data than drones originally are able to provide. For the sake of compliance, we have to refrain from using these methods, and are considering Phantom RTK drones.

Look forward to hearing from you!

Stunning shots

What are you hoping to be able to do?
It doesn't sound like conventional photogrammetry, which involved creating a mosaic of overlapping images.
That's not going to work with moving subjects or where most of the image is just water.

Wow, that's amazing shots.

So privileged to be able to see and take images of such beautiful creatures. Why oh why do we insist on harming them. Please do keep up to good work with them.

Great captures!

Wow, amazing captures

What Labroides said is relevant- unless your subjects stay still for the duration of you image collection, (and although I am one, you don't need to be a marine scientist to know they do not) photogrammetry will not work well or at all. The underlying assumption for photogrammetry to work is that things don't move around in between your overlapping image captures. Also as Labroides pointed out, water bad.

That being said, to the extent that you are able to image entire organisms (or at least the portion of interest), you can obtain very accurate measurements from single images. You may even be able to collect a few images over a single or a couple animals at a time and mosaic them together, while maintaining a consistent Grounds Sampling Distance (GSD or pixel size), but stopping short of true photogrammetry. At the altitudes shown in your captures, you're likely looking at  GSDs in the cm to mm range.And while  in our experience XY  accuracy of photomosaic and DSM products for the P4P RTK is indeed in the 1-3 cm range, I'm not sure how much that should matter for you. If you're mosaicking multiple images together it will help with the registration. The improved Z (altitude) accuracy of the RTK may be helpful for determining GSD as well.

And yes, very nice captures!

Hello and good day Strannik Wolkov. Thank you for sharing these information and these photos with us. Great work and thank you for your valued support.

embayweather Posted at 4-23 03:39
So privileged to be able to see and take images of such beautiful creatures. Why oh why do we insist on harming them. Please do keep up to good work with them.

Why oh why do we insist on harming them?
Who's insisting on harming them?
Commercial whaling dropped drastically from the mid-1960s and there hasa been a total moratorium in place since 1985/6.
https://iwc.int/commercial

Labroides Posted at 4-23 18:44
Why oh why do we insist on harming them?
Who's insisting on harming them?
Commercial whaling dropped drastically from the mid-1960s and there hasa been a total moratorium in place since 1985/6.

Countries are now, and have been, ignoring that moratorium using various excuses. We pump chemicals and plastic into their water, and leave behind various garbage like fishing nets. Perhaps insisting might be the wrong word, but we are doing very little not to harm them.

A J Posted at 4-22 23:26
Stunning shots

Thank you!

Strannik Wolkov Posted at 4-24 05:22
Thank you!

You're welcome

Wow, stunning shot

Dear All,  
Thank you for your reactions and feedback!  I think I have created some confusion by explicitly using the term photogrammetry. My apologies. Most simply described, photogrammetry refers to the science of making measurements from photographs, and photogrammetric analysis may be applied to one photograph or, as also mentioned by Labroides and Patiam , to a mosaic of overlapping images.

So what am I hoping to be able to do? I basically need one good photo of a whale that is surfacing with its body stretched and maximal exposed, with a minimum of water covering parts of the body as possible (see photo included).


To be able to calculate  and determine length and width of certain segments of the whale’s body,  exact height data / ground distance  is required.
Photos can be taken from max. 20-30 meter height – perhaps a bit lower but not too low – to avoid the effects of lens distortion on the photographs.

For that reason I am particularly interested in knowing how exact the Phantom RTK  is in its height measurements above sea level.

And yes, I fully agree with Embayweather, the Oceans and their sentient inhabitants need our full support and attention. For many of us UAVs have become a very useful tool – in my work a non-invasive way to study whales and hopefully contributing to their conservation.

@Strannik Wolkov-
Sounds like you have a good understanding of the capabilities and limitations of the system and the best approach. Yes, potentially the RTK's improved vertical accuracy could be helpful in improving the accuracy and precision of your measurements. I know the ultimate goal is to be non-invasive, but you'll want to groundtruth enough aerial measurements with in situ ones until you're confident the method works to the desired tolerance.

To help verify scale you could use the method we use underwater and add some small lightweight paired parallel lasers to your setup, but they'd have to be bright enough to see against the skin of the animal in daylight but still eye-safe, and they'll only give you scale in the plane onto which they are projected.
Good luck and keep us posted! Is your research exclusively on E. robustus, or do you study other species as well?

@ Patiam-
I am just starting to understand the possibilities of the P RTK.  Have been reading  some of your posts regarding this drone and you are definitely more experienced and competent in assessing its possibilities and technical reach. Since 2016, I have been doing  photoID flights and documenting behavior (video) between mother and calves and young calves of gray whales. We work in two teams covering  about 120 km shore line of a near-shore feeding area at the NE coast of Sakhalin in the Russian Far East. Similar programs are run in Australia, US-Hawaii and Mexico,  where cetaceans are observed in lagoons (i.e. LSI - OL gray whales  Baja California), the near-shore area and from boat-launched drones (southern right whales, humpbacks etc.) .

The RTK is completely new to me and I am trying to catch up. Besides my question on how accurate the vertical readings are, I also wanted to know if the separate available ground station is a required/recommended necessity to improve accuracy. So if I can pick your brain on these things that would be great.

Your suggestion for scaling is as genius as simple and provides extra data for verification and accuracy. Many thanks for that idea! Probably green laser and some tests are required to establish visibility of the laser at a certain height vs. lens distortion at that height.

@Strannik-
Glad to help a fellow marine scientist with their research!
For high-accuracy GPS on a drone you basically have 3 options:

• RTK using a local base you set up that brodcasts corrections to your mobile rover (aircraft in this case). This is what you would use the DJI D-RTK 2 base station for. NOTE: it would need to be on land or an unmoving offshore structure (not a boat), as it can't move at all during a particular data collection session once it is set up. Also, it works best if placed on a location for which you have already have high-accuracy coordinates. If using an established benchmark, those coordinates may be aavailable from the agency that installed the benchmark, or one can detrmine them using surveying methods (a long occupation with an RTK-capable GPS can use PPK to solve the corrdinates).
• RTK using a VRS (Virtual Reference Station) accessed via radio or the internet. Requires no local base station, but you must have connectivity to the VRS, and it should be nearby (~10-15km) if it is a single station. Services exist that use a network of surrounding stations to create a VRS right at your location as well.
• PPK (Post-Processed Kinnematic), which really isnt RTK at all (the RT stands for "Real Time") but with the right hardware and software can provide as good or better results after the fact (as the name implies). No base station is required as long as there is a nearby CORS (Continually Operating Reference Station - may of these are also VRS's) from which you can download data for the period you flew. Again, the important thing to remember here is that you will not enjoy high positional accuracy in real time, only after the flight is over. Depending on your requirements, this may or may not be OK.
• (There are other ways such as satellite-based correction subscriptions but implementing them on a COS DJI UAS is probably not feasible).
  

Regarding the P4PRTK, the jury is still out on whether it performs up to the advertised specs. In our experience, it does not. I'll post more about that later. The short story is that for photogrammetric mapping, the resulting orthomosaics and DSMs have great (2-3cm) XY accuracy & precision, but in Z it's no better than 50cm. Note that this is for the processed photogrammetry products, not necessarily the instantaneous accuracy of the AC positioning. But something is not working right in the system at this time.

As for lasers, yes green would probably be the way to go. They need to be parallel and as close to orthagonal to the CMOS plane as possible (at least along their axis). I saw in another thread you were looking into laser rangefinders. Good call, that's another possbile way for you to get your distance above the animal so you can determine scale. Bottom line (as you know), the P4's barometric sensor is too noisy and inaccurate for your needs.

Patiam greetings!  Thank you very much for your feedback and valuable advise. I have been prepping for the field season and was out for2 weeks . Received 2 sets of P4R's and D-RTK-2 ground stations. Did some height tests (see photos) . Which programs you recommend for accessing the metadata - e.g. height?

Here is a little write I did to help folks process the PPK data from a flight.

The original write-up is from Aerotas (link provided) and then I added additional images to explain a little more. Once you get the output CSV file, you can use that to correct the images GPS location based off the CORS and Satellite data.

I still need to do a write-up on how to use the CVS file to update the photos with the corrective data.

https://forum.dronedeploy.com/t/ ... g-with-rtklib/13217

Very nice shots!

LYVFD8142 Posted at 5-16 04:40
Here is a little write I did to help folks process the PPK data from a flight.

The original write-up is from Aerotas (link provided) and then I added additional images to explain a little more. Once you get the output CSV file, you can use that to correct the images GPS location based off the CORS and Satellite data.

LYVFD8142 good morning!   Thank you very much - am going to look at this right away! Much appreciated!