Hi,
I have a drone Mapping job coming up in the city area of downtown Boston Massachusetts USA requiring some GCPs. It is a small job but it is going to require some thinking and planning on my part. I would like to ask any operators out there about best techniques and practices of making and taking GCPs in a city setting. As an example what is the best practice to minimize the effects of buildings blocking or reflecting GNSS signals and yet maintaining an optimal layout for the GCPs? How do deal with and minimize problems with people taking or disturbing my equipment. Are there any best settings for city environment with an Emlid RS2+? I will be on some busy streets so making a GCP on the street itself may be impossible. Any caveats that people have come across operating in the city environment that you came across on the job that were not readily apparent? Thank you for any input and thoughts!
I’m using Metashape but I think the workflow should apply in general.
After the first stage of photo alignment I have the option to do a rigid transformation over control points. I also have the option to adjust camera parameters but I leave it to a later stage.
In urban areas there is plenty of reference points, I simply select something visible from the drone and in reasonably open areas. Manholes, road stripes, anything visible is doable.
I also measure 3-4 points that are not mistakable and evenly distributed (checkerboards, cross signs made with spray).
At this point I do a first rough registration over 3 unmistakable evenly distributed marks.
After that all the other control points are close to the feature, and I register them (with filter photos by markers I see the calculated position).
The trick is if you mark a manhole you cannot be sure which manhole is, but if you perform a first registration with only 3 (well known) points the error between control point and physical point will be in the 20-30 cm range, so you cannot be wrong.
If you add control points this way you refine the registration, and you can also update camera parameters calibration.
The process is very fast and easy, and in a urban environment solutions like checkerboards and similar are not doable.
Massimo
For Reach, as for any GNSS receiver, it may be challenging to obtain a stable FIX solution in urban areas, as it depends on the environmental conditions. But here are some tips for collecting the GCPs reliably.
Collect the points with Averaging instead of Instant collection. It’ll provide Reach with more data to calculate its position accurately.
The PDOP value should be under 2. You can monitor it on the Status screen in Emlid Flow.
The more satellites you have with a high SNR value, the better. Please ensure there are at least 5-7 satellites with an SNR greater than 35.
I hope it helps. But please feel free to share any experience from your project!
Be prepared to post process your control marks. If doing radio RTK make sure you log enough time for post processing. A minimum of 10 minutes, more if control marks are near buildings. Base receiver should be placed with no obstructions, perhaps on a roof if possible and also logging data. Keep your base-rover baselines at a minimum. This will help post processing if enough time is logged at rover. Another static receiver wouldn’t hurt either with a baseline at least as long as your project width. Logging data at all three receivers will give you confidence in your post processing results, IF you are using commercial post processing software as you will have three baselines for each point. This will give you a true mathematical error of your rover points unlike single baseline processors.
As for targets, there are multiple items to use, paint stripe corners, sidewalk seams with a paint mark, manholes as @studiodemarchi.net mentioned, fire hydrants, valve covers, etc.
Be prepared to purchase a business license for the city you are in. I would highly recommend this as first time offenders may have to pay a hefty fine as well as a year long license to go along with it.
If you do all of the above right the first time, it will save you many headaches.
My first recommendation is the use what’s there. Manholes, joints in concrete and striping are all great markers and you don’t have to worry about maintaining the network.
As far as placement you can only do what you can do. Your flight pattern will need to be configured to provide the best views possible. You only need 5-6 images per point to accurately geolocate and that hasn’t been much of an issue in my experience.
Regarding GNSS settings, I dial up my elevation mask to mitigate the “dirty” satellites on the horizon and more elevated paths that may be seriously hindered by structures. Last one I did had to be increased to 30% for me to get a fix.
Hey everybody, Thank you so much for your input. I truly appreciate you taking the time to answer my questions. Just wondering any input about security? With all the “Drone Hysteria” lately I want to avoid any confrontation and theft. Again Thank so much for your thoughts and time.
AirTag or similar in all your cases. Don’t let your base get out of your line of site (another reason to run local NTRIP) and hold your ground against drone Karens. You know the regulations and most of the public does not. Always have your registration and certificate available and be prepared to teach.
Hi, back from my epic learning adventure. I thought I would provide some feedback on my results for your comments and critique. So, after doing some research about GNSS receiver security, I found a product that would at least provide a sense of situational awareness for your receiver. About a year ago adopted an abused Beagle for our family pet. She was frightened of everything and would take off in the blink of a heartbeat. We bought a rechargeable tracker for her. From looking at the abilities of this tracker, I thought about mounting the tracker on the tripod so that you could get a notification if someone was moving your equipment. It has an alert radius that can be set up so that you get a notice on your phone if the tracker moves outside this area. It also uses a GPS mapping system and cell service to connect to its app. I know that this won’t replace a human monitoring your equipment but at least it will give you a heads up. You still have the option to tether a trip wire to an IED and then to your receiver. It is, however, a wash as you lose your receiver, but you also get the bad guy.
As far as trying to establish GCPs in the City jungle, it was a S-show. I tried to establish points that had the most amount of sky visible with hard objects on the ground, i.e… sewer covers, sidewalk expansion joints, etc. I used NTRIP with a Virtual Mount Point though cell service. The corrections came in correctly but, I had a heck of a time trying to establish a point on the ground in between buildings. I sampled 2 minutes to insure I was getting a good sample. There were multiple times the points that would drift in and out of float back to single and I would have to start the timing all over again ad nauseum. I shot more than the normal amount of points hoping that I could get some good data. It turned out about 1/3 of points were Floated.
The GPS coverage in my flight area on the ground was horrible. I am using an M3E with RTK. When I tried to get RTK lock for takeoff it kept giving the a position error move to better position warning. When I looked at the RTK tab in Pilot 2, I was seeing Fix and Float cycling back and forth on the M3E column. I moved take off position all around inside my flight area trying to get an RTK lock to no avail. I finally started to look for a takeoff area outside my flight area that would give me better GPS visibility. I found a spot 2 1/2 blocks away from my flight area that I could get RTK lock for TO. It was a little hectic crossing the busy streets while trying to keep the bird insight between the buildings and away from pedestrians to get me back to the flight area. The flight went ok except when I had to do a battery change and had to repeat the TO procedure again.
I recorded the GCPs through the Emlid Flow app survey page. I could see from results on the map that a lot of GCPs were off by looking at the building and street displays and its relation to the GCP’s position. Because I recorded so many points, I was able to delete some of the worst ones from processing.
I used PIX4D to do my processing. I did my Initial processing by incorporating all points using 3 images each for each point. Once the Initial processing was complete, I could see from the point cloud and quality report which GCP points were troublesome, and I then went into the GCP/MTP manager and deleted those points from the data set. I then did Initial processed again. This gave me better results but still not as good as one would like.
Questions:
Is there a hierarchy for RTK correction accuracy? What I mean is there a list of which technique you would use first in the order of correction accuracy then what technique to do next if you can’t do that one?
What things can I adjust through the software to get better results from the receiver in the city?
Would it be better to use my own base receiving NTRIP corrections and a rover getting corrections from my base?
What about using Emlid Caster system?
Is there anything else I could of thought of to give me better results?
Thanks once again for reading your way through my questions and replying!
Just trying to learn!
Thanks,
Wow, that’s a big paragraph to unpack but I’ll give you some quick notes. Downtown is just the worst and honestly, it’s better to just use a TS. Even when you see a fix I would be very leery. It can bounce to float any second so make sure you have your capture solution set to fix only.
Raising your elevation mask helps. Standard is about 15d so try dialing it up to 25-30.
Corrections from a local base via radio (LoRa) mounted high instead of NTRIP would help as long as you aren’t completely blocked by multiple structures. NTRIP just adds another element for interference.
When your M3E does that try turning it off RTK, getting to mission altitude and then turn it back on.
As for the security, not for this scenario, but think about a ring camera with a source of internet. I ran one off a Mi-Fi when I had to put my base in a sketchy place and would receive alerts if something came by. It was hidden in a tree. Now imagine walking up on a base and an alarm goes off or someone starts yelling at you, but don’t know where they are.
Being in the city, it’s going to be multi-path hell, no matter what. Good effing luck. Prob best to spread your control points further outside the area of interest that have clear sky view away from skyscrapers etc.
If u use a BASE or total station, you better get a helper and a ferocious dog to keep bumbs, thieves away or your stuff will grow legs and walk away real fast to the nearest pawn shop. Or better yet, just deliberately smashed for the hell of it.
Check out below for maybe permanent reference marks to tie into, data sheets etc in your area and/or get with the city:
I agree with @timd1971. It might be more challenging to achieve a stable and reliable fix in a dense urban area. The receiver is highly likely to suffer from cycle slips and multipath.
I think putting the GCPs outside the area of interest is a nice workaround. You can try adjusting your survey boundary and establishing the GCPs in an area with a clear sky view.
Alternatively, you might find some geodetic control points in the city with good visibility in the images. These could serve as GCPs, but the challenge will be accurately marking them in your photogrammetry software.
