That is impressive and not what I've seen, the ground control targets we've set for drone surveys are always very extensive. Time consuming and dense. Mostly I could have driven a 4-wheeler around the site in less time than flying it and controlling it.?ÿ
I've also tested the Delair UX11.?ÿ While not quite as accurate as the Wingtra, it was still pretty good without GCPs.?ÿ It was within 0.2' vertically.
The new generation of PPK drones are quite impressive.?ÿ I wish they would add a high-precision IMU for areas that are difficult to calibrate (trees, water, etc.).
I think EVERYONE would like a good IMU to assist in the processing.?ÿ They are still pricey, and ITARS doesn't want anyone to have easy access to them for nefarious reasons too I suppose.
All depends on the site and if you can drive it. Some of the quarries it used to take me all day to walk can be done in less than an hours flight (repeat jobs have permanent ground boards in place). Typical board spacing we use is around 300 metres, or three across the width if that is less.
Last weeks job was 4.5km long, 0.4m wide - flying time around 3 hours. It took a day and a half on site just to add in the GPS check points the client wanted. This was a set of fields under crop (kale, broccoli, cabbage, cauliflower) so there was no vehicle access apart from a single track. Job is looking for ground settlement from previous utility works. Comparison with last year was very impressive; all the small twists and "jags" in last years contours (0.1m), which we thought might have been errors were replicated in this years results.
Whoops - meant 0.4KM!
?ÿ
?ÿ
Wow so much knowledge and experience here. The first thing besides the 5? s5 that caught my eye is the multitrack prism. That is your weakest link especially in active mode if any vertical change in site exist. ?ÿThat target is less accurate than what your s5 5? angular accuracy is. ?ÿI would like to ask this question though. ?ÿIs it accuracy or relative precision on that site your truly trying to get down to. How well things check across the site. For example you could locate everything with good rtk measurements at-least two observations at completely different times of day. It will take more than 2 direct and reverse observations for a 5? gun to get you down across 40 acres. ?ÿGet a traverse kit and up the rounds to 6 or 8. Use force centering techniques carefully setting up the main traverse with the ability to turn to other points from multiple setups. Build good strength of figure. Allow the rtk observations to move weighting as much as you can afford to restrict weighting of your total station. ?ÿUse all observations in TBC. Your s5 strength will be the distance as it is 1mm + 2ppm I believe check me on that. ?ÿBut keep that in mind as you set up your network. ?ÿA regular traverse is not 100% necessary as we do them think more network or triangulation techniques but remember your angle accuracy is 5? thats with 2 direct and reverse to achieve that. The multi track is a roughly 10? siting prism by itself. On a pole so weakest link. The Trimble traverse kit is like 1750$. I would have a couple few sets so i could turn angle to more than just a back-site and foresight . ?ÿI would have triangles built all over redundancy redundancy redundancy. ?ÿCross ties everywhere I could. ?ÿIf you moved to a 1 sec instrument the old standards for 1st second 3rd order work would be a great read. ?ÿIt could help you plan. You will have to do 3 to 4 times the work with the tools you have to prove your results. ?ÿIt can be done but lots of good procedures and understanding of measurements will be key and the error of equipment itself. ?ÿOh the multi track might be 5? sighting. I can??t remember. But i did a network not long ago and used the multitrack on a point in center of a project for a sanity check. ?ÿJust left it there as i traversed around a site. ?ÿIts not a very accurate piece of equipment for distance or angles especially vertical and i used it on semi active. ?ÿ
You are forgetting this is to look at the results from a Wingtra. At best the accuracies from the photography are going to be only to a couple of centimetres, so the ground survey checks need only to be to around 1cm. Simple ground survey will do for the network, nothing fancy. No benefit in turning a couple of hours job into a weekend.
Our later GPS RTK units blow pass any accuracy requirements for photogrammetry targets. In the old days it was trig levels and T2/distance meters, today it's GNSS units and for checks a robot setup which normally will find control in the 5mm-10mm range. Far excedding the accuracy needs.?ÿ
Actually, even for tight control GNSS units are beginning to be our go-to. I just finished control for a site survey for a PUD that will be condos and storage units. We set 4 control points around the site with GPS, occupied them with the robot and found .02' vertically and the same horizontally. We were finished at that point, held the RTK numbers fixed and will never look back. I'm beginning to see most error in our new equipment is measure up/rod bubbles/optical plummets and not much else.?ÿ
Make the NGS benchmark one of your Targets.
John, you're the last guy I'd expect to see writing "bench mark" as one word.
Jim, I too make mistakes; Don't YOU? ?????ÿ
Thanks Jim I guess I better check my spelling more closely from now on.
?ÿ
Jim, I too make mistakes; Don't YOU?
Yesterday I dropped a chuck key on a dial indicator face - does that count?
@jim-frame?ÿ ?ÿNO
Make the NGS benchmark one of your Targets.
John, you're the last guy I'd expect to see writing "bench mark" as one word.
Good on you, Jim. One of my pet peeves. Even NGS violates this in various documents. If you want to test the speed of your computer, then it's a benchmark. If you are talking about a physical object in the ground (or vertically mounted on a structure) that is referenced to a datum, it is a bench mark.
@chris-mills oh man I thought. ?ÿ .02 hundredths of a foot he was trying to achieve if its 2cm degnabit just rtk the jokers in and still do ar minimum two observations at different times no traverse needed whatsoever. ?ÿThats what i would do. Come hit them all one morning go eat lunch have a beer or two comeback that afternoon hit them again since it??s Trimble average right in the dc. ?ÿUnless a large outlier. ?ÿSo sorry. I was way over kill. ?ÿLol. ?ÿWhy can??t we all just pick a unit and work in it. Meter survey feet international feet inches chains rods vera geezers. ?ÿ Lets make a new unit and just say from now own thats it. ?ÿMy bad. Thank you for catching my ignorance. ?ÿ?ÿ
Good on you, Jim.
I've been as guilty as the next guy, but in recent years I've been trying to conform to the NGS preferred format.?ÿ Mostly as a result of John's needling!
@chris-mills?ÿ The original poster was working on a Master's degree. He says "in my case, the Sony RX1RII camera can achieve a, GSD of 0.025 ft agl. See your post above at 6:40am where you say 2 cm (photo) and the ground survey check only need to be 1 cm.?ÿ 0.025 ft = ~7.6mm ;?ÿ 1cm = 10mm?ÿ ?ÿ For a test on the accuracy of the Wingtra you need to have ground control 10 times better than what you think the Wingtra is capable of. I would think + or -?ÿ 1mm would do on ground control for x and y coord. and + or - 2mm in the vertical.
I appreciate the point you are making in terms of theroetically what is needed for research, but there are so many other factors affecting the accuracy of the derived information (such as the factors applied to allow for vegetation height - even grass if it has been walked on or mown) that a full degree of magnitude isn't really required. Note also that the targets are described as having a fixing hole in the middle. If you are looking for 1mm accuracy you need a fixing at each corner so that there is no slope on the target. Depending upon the software used there can also be errors developed close to the edge of a peg fixed target as the ground and edge heights of the board are not the same - not great but well over the 1mm. threshold. In that respect large targets are better as the centre of the target area will give a true figure to the photogrammtrical output.
Sorry, I missed replying to one of your points. Resolution isn't the same as accuracy (as you well know). If the resolution from the Sony is only equivalent to 8mm. then the accuracy obtained is likely to be 2 to 3 times that figure.
Hey fellas, sorry I didn't report back.?ÿ Here is a link to download my master's report, which outlines the procedures I utilized:
Milanes Master's Report - UC Denver Fall 2021
I was both pleased and disappointed with the results from my project.?ÿ The location I chose was not ideal for this type of testing.?ÿ I needed a larger area, with better visual texture, and more flights to really differentiate between the flight parameters.
However, even with a sub-optimal test site, the Wingtra was still able to deliver horizontal accuracies of around 0.05' and vertical around 0.15', which is within ASPRS of 1-2xGSD horizontal and 2-3xGSD vertical, without ground control.?ÿ The Wingtra is an impressive UAV surveying platform.
Let me know if you have any questions about my project.