Welcome, Commercial Drone Pilots!
Join our growing community today!
Sign up

Do construction jobs need centimeter level accuracy?

The Phantom 4 Pro V2 is what I use for mapping and its mechanical shutter, 20MP, 1" sensor perform very well especially with dialed in Manual settings. I have a custom aftermarket PPK kit on the drone and also will use both PPK and GCPs but also just GCPs. My map with the best reported accuracy by comparison to control points was a small one with 5 GCPs and 5 Checkpoints, no PPK corrected images and the results were excellent. Actually I cannot state the mm level RMS error as in the X and Y horizontal it is slightly below GSD and accuracy cannot be under GSD.

View attachment 3277


The Autel Evo 2 RTK or Enterprise with RTK module is a very interesting new solution. You are basically only NTRIP/VRS (Internet corrections) or a known point with an Emlid Reach RS2 (Also can act as a caster to send out its own NTRIP) away from adding GNSS accuracy.
It has a 1" sensor, long flight times, the ability to swap camera payloads (The connection is not meant to swap payloads an infinite number of times in my opinion) and a stated cm precision in both RTK and PPK workflows, with the PPK workflow using a produced RINEX file.
A sister forum member JMason has been conducting some really interesting tests with this unit and the results have been very good. I am in no way an image expert, but I have read that Autel does a much better job in converting images to JPEG format and that many people claim that this helps with more points in your photogrammetry derived point clouds. It does lack a mechanical shutter which would enable you to conduct missions at a higher speed and help prevent the roller shutter effect, but the desktop applications like Pix4D and Agi Meta both have algorithms that can deal with it. You could also fly at a lower speed, but I would personally like to read more about others experiences and tests.

Link to Evo 2 RTK tests on sister forum. He has several threads on different tests, including thermal mapping and is a wealth of information on this new drone with his tests that he posts for all members to view.
Wow that looks very accurate, very small errors. I was looking at the phantom 4 RTK or the Autel Evo II but will likely start with a standard phantom 4 pro with no RTK to keep costs down. Is it challenging measuring accurate GCP and control point coordinates or is that something which can be learned with the proper GNSS equipment?
 
For a non-RTK drone less than $1800 the Phantom 4 Pro is still the best mapper you can get. Just stay away from the "+" models. If you intend to continue with mapping and get deeper into accuracies you will want to make your way to an RTK drone sooner rather than later. Or you'll need to invest in some GNSS equipment. I would keep the brand in mind whether it be DJI or Autel. Things like batteries, chargers and props will all cross over. Personally we fly the Yuneec H520E RTK which is a slightly larger hexacopter but I would still choose the Autel Evo II Pro RTK over the Phantom 4 RTK although they are hard to come by and Autel could not give us a definite answer as to when they will be available again. Poor timing right before the holidays...
You haven't thought about the Evo 2 Enterprise with the RTK module? It offers so much more value than the RTK package.

The RTK package from what I saw on the only US vendor I could find wasn't comparable.

Enterprise Kit

(1) Aircraft w/ Pro 1" 6K Camera & Gimbal Cover
(3) Flight Batteries
(1) 7.9” Smart Controller
(1) 110V Wall Charger
(1) Smart Controller Charger
(1) Smart Controller Charging Cable
(1) Speaker
(1) Search Light
(1) Strobe
(1) Chest Strap
(1) Attachment Port Cover
(3) Propeller Pairs
(1) Hard Rugged Case


RTK Kit

1x EVO II Pro RTK Drone
1x RTK Module
1x Durable had transport case
1x Wall charger
1x EVO II battery

Price difference was about $300 more for the Enterprise.
 
Thanks. I think I will just go with the phantom 4 pro without RTK and upgrade from there. I am trying to keep my start-up costs as low as possible to see if I enjoy mapping and if there is a market for it in my area. Worst case I can keep the phantom 4 pro for regular photo/video work.

In terms of GNSS equipment it seems like it may be cheaper to get a phantom 4 pro without RTK and an Emlid reach RS2 compared to a autel evo II Pro RTK with the base station. Alternatively I can probably subcontract the work out to a surveyor on my early jobs before investing in GNSS equipment.

I also think I will likely have to upgrade my computer setup for processing if I do it locally and buy software for mapping such as Agisoft Metashape or a Pix4D subscription.
I think a non-RTK P4P is a great choice and I would share that it would be a good idea to look at a PPK setup and if the budget allows. The nice part is that you can scale as you want. As for photogrammetry software either of those are great choices. Depending on what you are doing with the drone and how often you will figure out if it is feasible to locally process.
 
  • Like
Reactions: Ask Ketchum
Thanks for the tips, I will definitely spend some time studying GIS, doing practice flights and networking before going out to market my services. Is a Phantom 4 Pro a good drone to get started?
If you are flying 5-7 acre sites sure. But there can be problems operating in controlled airspace where the DJI version and the FAA version differ. That may require a manual unlock of airspace from DJI, which has always bothered me. If you are flying larger sites then I would suggest that you find something more capable and not Chinese in origin; especially if you are trying for state or federal contracts.
 
I also think I will likely have to upgrade my computer setup for processing if I do it locally and buy software for mapping such as Agisoft Metashape or a Pix4D subscription.
Before you upgrade any hardware decide on which software package you are going to use. Then upgrade the hardware to meet the software requirements. Also, how many images do you think on average you are going to be stitching? For large numbers per job you are going to be better off buying a dedicated machine to run your processing. On my old Dell workstation, I had a 5500 image job that ran for 8 days before I finally shut it down. Replaced the Dell with an Alienware Threadripper and cut processing time down to 38 hours for the same job.
My point being you need a dedicated machine or multiple machines because you aren't going without a computer to process a job right?
 
Before you upgrade any hardware decide on which software package you are going to use. Then upgrade the hardware to meet the software requirements. Also, how many images do you think on average you are going to be stitching? For large numbers per job you are going to be better off buying a dedicated machine to run your processing. On my old Dell workstation, I had a 5500 image job that ran for 8 days before I finally shut it down. Replaced the Dell with an Alienware Threadripper and cut processing time down to 38 hours for the same job.
My point being you need a dedicated machine or multiple machines because you aren't going without a computer to process a job right?
Different processing softwares don't need different hardware so it's more basically the fastest processor you can find with a graphics card with as many CUDA cores as you can afford. Matched RAM is also a big factor. This is why allot of companies that want to process their own data "locally" have started doing it on a virtual workstation(s). I even have a big rig at my house that I can remote into if needed.
 
  • Like
Reactions: Ask Ketchum
If you are flying 5-7 acre sites sure. But there can be problems operating in controlled airspace where the DJI version and the FAA version differ. That may require a manual unlock of airspace from DJI, which has always bothered me. If you are flying larger sites then I would suggest that you find something more capable and not Chinese in origin; especially if you are trying for state or federal contracts.
Thanks, I'm in Canada so not sure that applies but good to keep in mind.
 
I think a non-RTK P4P is a great choice and I would share that it would be a good idea to look at a PPK setup and if the budget allows. The nice part is that you can scale as you want. As for photogrammetry software either of those are great choices. Depending on what you are doing with the drone and how often you will figure out if it is feasible to locally process.
Thanks, in terms of PPK vs RTK which one would be better? From my research it seems like PPK does corrections during processing time while RTK would do corrections in flight. It also sounds like both PPK and RTK would require a ground based base station. Would PPK and RTK remove the need for GCPs or would they still be needed? From the looks of it, the cheapest solution may be to utilize a non-RTK P4P with a PPK upgrade kit installed on it and some sort of ground station.
 
  • Like
Reactions: adm_geomatics
Thanks, in terms of PPK vs RTK which one would be better? From my research it seems like PPK does corrections during processing time while RTK would do corrections in flight. It also sounds like both PPK and RTK would require a ground based base station. Would PPK and RTK remove the need for GCPs or would they still be needed? From the looks of it, the cheapest solution may be to utilize a non-RTK P4P with a PPK upgrade kit installed on it and some sort of ground station.
PPK is better simply for the fact that you have control of every aspect of the observations after the fact. We use PPK for every certified mission meaning that it has to be signed off on legally. I am much more comfortable with RTK on a drone now than I was 2 years ago because the technology, particularly software, has come so far in that short amount of time. We still check RTK results with PPK but only rewrite the geotags if there is a significant discrepancy.

You don't necessarily need ground control points but you still definitely need at least three checkpoints for your processing. The scenario that we are in with construction requires ground control points every time just because of the local coordinate systems that come into play. As CAD files get better at being on a true grid then the need for gcps will continue to decline but there will always be a need for at least vertical checks before going into any other software.

I think taking the little extra effort to learn the PPK process will be a double benefit when flying with a standard Phantom 4 Pro. Other reason being that you are going to have much more support with both flight software and the community.
 
  • Like
Reactions: Ask Ketchum

Members online

Forum statistics

Threads
4,277
Messages
37,605
Members
5,969
Latest member
KC5JIM