Has anyone use the Topodrone PPK kit for the Mavic 2 Pro with any success? Is this worth the investment to get into survey work for relatively small construction project surveys. Link TOPODRONE DJI Mavic 2 Pro PPK
Topodrone PPK Kit
- Thread starter Salty Sam
- Start date
That looks like a nice system... I don't know much about that specific brand & kit. If you had a Mavic 2, it'd probably be one of the few options to upgrade a Mavic to PPK.
I'm focusing on the PPK / RTK tangent for construction, Ag, and 3D Modeling. I've bounced around several products and various
Platforms. Not as much focus is emphasized as previously, but the camera & shutter remain a variable and rolling shutters aren't perfered over mechanical shutter and larger sensor camera. That is one reason the Phantom 4Pro remains popular... the camera and mechanical shutter under 1/2000. Although I have all Mavics and P4P... and I do use the M2P for some progress photos on construction, I normally use the Inspire2 with X4S. The X4S is exact same camera as P4P.
A PPK company I've recently discovered looks great to me.
Called BAAM Tech:
What Is A PPK System? Aerial Mapping Drone With GPS Equipment: BAAM
At BAAM Tech, we created a custom PPK system designed to make your aerial mapping projects easier. Call us for more information!
baam.tech
They use a PPK system that can be moved from Base to Airframe... so only buying 1 unit. They offer an optional additional Rover for GPS and GCP in similar removable operation to easily move unit from Rover tripod, etc. Their accuracy is top notch.
The simplicity of package certainly doesn't minimize the quality or accuracy of the system.
I'm looking mainly at the Inspire 2 / M210 with X4S over P4P, and they don't offer a M2P kit.
They modify the camera to be calibrated & in connection with the PPK.
I'll look a bit more at the TopoDrone, looks like a nice package; although the M2P's rolling shutter is less appealing. There is software adjustment within most mapping apps, but it's still not as accurate as mechanical shutter.
adm_geomatics
Well-Known Member
- Joined
- Nov 30, 2019
- Messages
- 507
- Reaction score
- 277
- Age
- 50
- Location
- Leander, Texas
- Website
- www.admgeomatics.com
This is a very good solution. You can pair it with a GNSS receiver like the Emlid Reach RS2 and post process with their Toposetter software. It is only $200/yr for a perpetual license. Or you can use RTKLIB and Geosetter for free if you don't mind a little bit more manual workflow. The $200 is well worth it though. We also have an NTRIP subscription and can use the RS2 to set GCP's when it's not being used as a base for the drone. This also allows you to set a hard globally accurate point for the base instead of just averaging something in.
adm_geomatics
Well-Known Member
- Joined
- Nov 30, 2019
- Messages
- 507
- Reaction score
- 277
- Age
- 50
- Location
- Leander, Texas
- Website
- www.admgeomatics.com
Not sure who you replied to, but I was referencing the Topodrone kit. BAAM does not have a direct solution for the M2P.
LOL Yes I went to the site and found that out. I was actually replying to Dougcjohn. I will keep looking for a solution to shoot sites of about 1 acre or less
adm_geomatics
Well-Known Member
- Joined
- Nov 30, 2019
- Messages
- 507
- Reaction score
- 277
- Age
- 50
- Location
- Leander, Texas
- Website
- www.admgeomatics.com
What is it that you are doing? For sites that small I don't think you are going to see enough gain to substantiate it. You'd be better off doing the stop and shoot method.
I am not a surveyor. I was considering this as a value added while doing construction photographs. It seems to me with a little added effort (and equipment) I could give the contractor progress elevations and possibly support survey firms that don't have photogrammetric capabilities. Might not be as easy as I thought but I am always looking for and edge when trying to sell my services.
adm_geomatics
Well-Known Member
- Joined
- Nov 30, 2019
- Messages
- 507
- Reaction score
- 277
- Age
- 50
- Location
- Leander, Texas
- Website
- www.admgeomatics.com
Glad I asked. Starting PPK to add services to your portfolio is going to be a bit different than if you were already providing those services on some level with or without the use of GCP's. The first thing I would recommend you brush up on is coordinate systems more specifically the difference between geodetic and cartesian systems. The transformation between these is what is going to happen to relate the information shot by the drone to pretty much any terrestrial control. Then you will want to understand what coordinate systems the surveyors and engineers in your area are using. If you are wanting to quantify stockpiles you can already do that without GCP's or PPK as long as they are not larger than an acre. If they get larger like you might see at a mine or quarry then you will want GCP's, but still don't need PPK. PPK comes into the scenario when you want to start comparing flight vs flight production or drone vs CAD data in which you will want to still use a few GCP's for ultimate accuracy. You can forego GCP's with PPK on sites of less than 10 acres as long as the same coordinate is used on the base station or you use the same corrections service every time. Just a heads-up of course as you progress forward.
Sorry if mislead... I had indicated BAAM lacked a M2P PPK kit, was providing other options to consider that use a mechanical camera.
Keep in mind, your 2D orthogramatry mosaic produced by your fly over, single or double grid does capture in centimeter accuracy ( advertised between 3-8cm) for Construction to perform measurements: length, area, mass... volumetric not so accurate. Adding GCP helps to further align the images, without GPS too.
The RTK / PPK does tighten that accuracy but the main benefit is accuracy to the physical ground & survey coordinates points.
Many Construction Mgmt enjoy the quick access and measurement tools within MapsMadeEasy. I often for smaller updates will peel off enough images to slid under the free 250pts of MME, and for other sets use greater images and CM accuracy (greater zoom detail & measurements).
The measurements of the slab, spacing of rods, distance from other sections are all very accurate... but on the aspect of the "image", not relative to the accuracy of ground points. To associate the image points to physical ground and measure accuracy to physical point on ground requires RTK / PPK or GCP measured with Rover.
For measuring the various build stage and various demonsions it doesn't really matter if the accurate "image" is accurately tied to the ground coordinates. A 25.5' Wall is a 25.5' wall, regardless if it's accurately aligned with the precise location on ground.
As non-surveyors, we can provide RTK/PPK ground associations, although we can't represent them as survey accurate, nor can the construction project utilize them as survey accurate... that requires a Survey Team.
adm_geomatics
Well-Known Member
- Joined
- Nov 30, 2019
- Messages
- 507
- Reaction score
- 277
- Age
- 50
- Location
- Leander, Texas
- Website
- www.admgeomatics.com
We get very accurate volumetrics with nadir only flights and terrain processing in DroneDeploy.
The main benefit to RTK/PPK augmented GNSS is the relativity factor. The entire map then becomes more accurate to itself, warping is all but eliminated and single-point calibrations can be made to get it close to 7-8cm stakeout accuracy without GCP's. The 2-3cm accuracy that everyone advertises is the relativity of the map. You have to have GCP's to get sub-5cm stakeout accuracy even if the base is on a known site control point. We won't even get into localization, but I will say that flying a drone off of unlocalized control on a construction site that has been localized is a big no-no.
Chasco
Thank you for your reply. This information is quite enlightening. I guess if it were easy everyone would be doing it. I will study up on our coordinate systems with the help of a local surveyor friend of mine that already does photogrammetric surveys. Looks like there is a lot for me to learn.
adm_geomatics
Well-Known Member
- Joined
- Nov 30, 2019
- Messages
- 507
- Reaction score
- 277
- Age
- 50
- Location
- Leander, Texas
- Website
- www.admgeomatics.com
Having a link to the survey community will give you a huge edge! I was lucky to come from survey and have seen many pilots/companies struggle over the last couple of years trying to do what we have been doing. They are catching up fast though! It's not hard at all, just different information that most people aren't exposed to. Kind of like the 107 test. It was easy, but the first time I looked at the sample data I was a little intimidated by things I have never heard of before.
Metashape does fairly well on volumetrics, as DD too.. MME not so good.
Without GCP or known material, volumetric accuracy isn't great based on NADIR only... small elevation inaccuracies can effect the measurement. Larger stockpiles maybe not so precise.
adm_geomatics
Well-Known Member
- Joined
- Nov 30, 2019
- Messages
- 507
- Reaction score
- 277
- Age
- 50
- Location
- Leander, Texas
- Website
- www.admgeomatics.com
I agree MME is decent for 2D Maps and 3D models just to look at. They use to do onscreen GCP's (manual tie-points) which were horrible, but finally introduced real GCP imports a little while ago. Their workflow is terrible and something is wrong with their use of the GCP's. I can run the same project and GCP's through all 3 and MME is easily 5% out of norm.
As for stockpiles and nadir-only we have done extensive testing using GPS and robotic terrestrial surveys and everything we have documented with the drone is within 3-5% of the stockpile volume per the trucking tickets which when trustworthy are the measure. The drone was closer than either of the standard survey methods each time. In construction material moves too fast to think that you are going to get much better live accuracy than that consistently. Like I said above a scenario like a mine or quarry that has 100,000cy+ stockpiles would benefit from GCP's, but a materials hauler or construction company with average 10,000cy piles not so much. Basically the larger than area gets the worse the accuracy gets percentage-wise exponentially because of warping. You can mitigate some of this by flying higher, but then you lose the fine detail so it's always a trade-off.
I am curious, how are you quantifying stockpiles for comparison against the drone data to say that the drone data is not accurate? I can only think of two ways so I am curious what your benchmark is.
Last edited:
MME 2D has improved, but their 3D is terrible other than a rough structure. Run the same images through Metashape (MS) or Reality Capture (RC) and the 3D looks night & day better... sometimes the MS 3D from nadir position looks about as good as MME 2D.
The one thing MME offers is ease of sharing the results and quick simplistic measuring. Sharing the URL and creating a Location to present multiple weeks that a Client can simply click and review each session, very nice! That is much more complex to present with any higher quality "local processing" application. I'd really like to see more "Image & 3D Players" to improve presentation.
That is very enlightening to read your volumetric test results... 3-5% tolerance is much better than I would have expected and at all nardir. I'm not skeptical on practice applied to obtain result of volumetric, more drone limitations. All nadir DEM are based on drones circuits for their accuracy of elevation.
Primarily due to the drones own specifications of it's accuracy & calibration tolerance of temp & weather; both GPS and Elevation circuitry is not highly accurate on consumer or prosumer drones... a lot of latitude in sensing precise changes and that carries over to the stockpile height and base accuracy. Obtaining either a more precise sensor accuracy via PPK & Base measurements or obtaining stockpile base & some pt of height above (not necessarily peak) provides more precise instrument to obtain calculations.
But your test results are encouraging!
The one thing MME offers is ease of sharing the results and quick simplistic measuring. Sharing the URL and creating a Location to present multiple weeks that a Client can simply click and review each session, very nice! That is much more complex to present with any higher quality "local processing" application. I'd really like to see more "Image & 3D Players" to improve presentation.
That is very enlightening to read your volumetric test results... 3-5% tolerance is much better than I would have expected and at all nardir. I'm not skeptical on practice applied to obtain result of volumetric, more drone limitations. All nadir DEM are based on drones circuits for their accuracy of elevation.
Primarily due to the drones own specifications of it's accuracy & calibration tolerance of temp & weather; both GPS and Elevation circuitry is not highly accurate on consumer or prosumer drones... a lot of latitude in sensing precise changes and that carries over to the stockpile height and base accuracy. Obtaining either a more precise sensor accuracy via PPK & Base measurements or obtaining stockpile base & some pt of height above (not necessarily peak) provides more precise instrument to obtain calculations.
But your test results are encouraging!
adm_geomatics
Well-Known Member
- Joined
- Nov 30, 2019
- Messages
- 507
- Reaction score
- 277
- Age
- 50
- Location
- Leander, Texas
- Website
- www.admgeomatics.com
I think some more research into the algorithms might put your mind at ease. Decent volumetrics can be had without geotags. All geotags do is place it at a scale and some place on the planet. Photogrammetry including aerial has been around allot longer than GNSS. Adding these geotags obviously enhances the process and the machine can exploit 1 image against 10-12 to do very precise bundle adjustments which greatly reduce the error from the figures you see on the manufacturer's spec sheet. When we are trying to distinguish between nadir and oblique imagery used in reconstruction we need to realize there really is no true nadir unless you are absolutely over the center of the point. All nadir imagery includes somewhat of an oblique angle. When the angles are taken to what we would usually consider oblique you are actually introducing more distortion and error into the processing. People think that oblique imagery is so much better just because it makes a structure or a building look better, but that is only because the angle from the camera to the face of the structure is less oblique... Think about that one for a minute.
Dave Pitman
Well-Known Member
I agree with Michael"s (@chasco ) assessment with regard to volumetrics. The amount of distortion that can occur in a small area like a typical stockpile is minimal and insignificant to a good result. Good overlap, enough views, and image sharpness is much more important. Any tying into a local CRS has no value. Also, in my testing, MME did a good job compared to the software I tested it against; Metashape and Correlator 3D.
This is true as well. I think MME is simply stretching and fitting the processed ortho rather than actually processing from scratch relationships based upon GCPs. In other words, they are just making it so subsequent orthos will somewhat visually stack but is not necessarily removing distortion or tying to any particular CRS.
Dave & Michael, thanks for the detail... all useful.
Doesn't that also state the need for purchasing and adding RTK / PPK isn't as needed for non-survey 2D Ortho imaging (and Modeling) as with the majority of construction projects. More applicable to use GCP without the need for RTK / PPK?
I've questioned the need to purchase PPK, felt it wasn't beneficial / economical for the project.
General discussion consensus seems to always indicate the need to purchase RTK platform, or add PPK kit to accomplish accurate mapping. I acknowledge you gain accuracy, although the above posts on volumetrics implies the increased accuracy isn't enough to alter volume.
Other than associating a GPS point to earth, what gains are RTK / PPK hardware and additional field time offering?
Doesn't that also state the need for purchasing and adding RTK / PPK isn't as needed for non-survey 2D Ortho imaging (and Modeling) as with the majority of construction projects. More applicable to use GCP without the need for RTK / PPK?
I've questioned the need to purchase PPK, felt it wasn't beneficial / economical for the project.
General discussion consensus seems to always indicate the need to purchase RTK platform, or add PPK kit to accomplish accurate mapping. I acknowledge you gain accuracy, although the above posts on volumetrics implies the increased accuracy isn't enough to alter volume.
Other than associating a GPS point to earth, what gains are RTK / PPK hardware and additional field time offering?
Last edited:
Dave Pitman
Well-Known Member
Doug, in short, GCPs, and drone based RTK and PPK serve the purpose of removing distortion in the resulting surface mesh and relating or tying the mesh to a known coordinate reference system (CRS).
If the area of interest is either small in scale (like individual stockpiles), or has no requirement to be tied to a CRS, then the extra work is not necessary. On the other hand, as the scale of the area grows, or there is a requirement for points on the ground to align with points on the ground as referenced to a CRS, then proper use of GCPs, and or RTK/PPK come into the picture.
So, for general construction progress orthos that are not used to calculate much of anything and are just used as a reference, the GCP workflow doesn't add much value. But, for example, if the surface mesh you are producing is going to be used by a grading contractor to see how much material they need to remove or haul in, then it absolutely needs to have a working ground control workflow that can be verified.
If the area of interest is either small in scale (like individual stockpiles), or has no requirement to be tied to a CRS, then the extra work is not necessary. On the other hand, as the scale of the area grows, or there is a requirement for points on the ground to align with points on the ground as referenced to a CRS, then proper use of GCPs, and or RTK/PPK come into the picture.
So, for general construction progress orthos that are not used to calculate much of anything and are just used as a reference, the GCP workflow doesn't add much value. But, for example, if the surface mesh you are producing is going to be used by a grading contractor to see how much material they need to remove or haul in, then it absolutely needs to have a working ground control workflow that can be verified.
New Posts
-
Sharing My Music and Sound Effects - Over 2000 Tracks- Latest: Eric Matyas
-
-
-
-
Members online
No members online now.