Activity Feed › Discussion Forums › Photogrammetry, LiDAR & UAS › Drone Spec for Surveys
Drone Spec for Surveys
Posted by Surveyor_BB on September 20, 2022 at 8:40 amHi Guys,
We’re thinking of getting a drone for our survey department and have been inundated with a wealth of information.
We plan on getting an RTK Drone and use one or our existing RTK licenses out of a spare GPS kit.
Would you advise using Ground control points still? I’m thinking just as a check
Is there any official guidance on the amount of ground control points needed for a survey? (spacing, amount etc)
Thanks
P
rover83 replied 1 year, 2 months ago 12 Members · 14 Replies- 14 Replies
Having some check points is always helpful. If you have large elevation relief, have a point up high and on at the low end.
Without Ground Control points there is no way to prove the data’s accuracy to the ground. You may have a great set of data but need to be able to prove its better than other data sets. My humble opinion….. We have about 13 pilots (2 pilots, 11 Part 107) and are doing 100+ UAV projects a year, and each and every project a PLS will review and oversee the capture area, GCP layout, control and QA/QC with final separate ground control confidence checkpoints. This is our typical workflow.
- Posted by: @jered-mcgrath-pls
Without Ground Control points there is no way to prove the data’s accuracy to the ground. You may have a great set of data but need to be able to prove its better than other data sets. My humble opinion….. We have about 13 pilots (2 pilots, 11 Part 107) and are doing 100+ UAV projects a year, and each and every project a PLS will review and oversee the capture area, GCP layout, control and QA/QC with final separate ground control confidence checkpoints. This is our typical workflow.
We’re just getting into scanning and drone work.
Would you know of any good links on procedures that I can forward to our people specifying what types and how many GCP are recommended for varying types of drone or scanning surveys?
On the question of GCP quantity and spacing… by far the two biggest companies in the drone mapping and data processing space (for large projects requiring accuracy) are Pix4D and DroneDeploy.
Pix4D published this study of GCP density from Nevada DOT (2018, but still very applicable in my opinion):
https://www.pix4d.com/blog/GCP-accuracy-drone-maps
“Effectively, this means that 11-13 GCPs would likely be the most efficient number for this site and geometry.” (34 acre site)
“Pix4D recommends using approximately 5 to 10 GCPs and indicates that ??more GCPs do not contribute significantly to increasing the accuracy.?
DroneDeploy has some recommendations on their site as well (overview of their company recommendation, no stats to back it up):
https://help.dronedeploy.com/hc/en-us/articles/1500004965282-Ground-Control-Points
“DroneDeploy requires a minimum of 4 GCPs and we suggest using 8 GCPs for maps up to 50 acres. For maps over 50 acres, we recommend adding at least 1 Checkpoint for every additional 10 acres. GCPs should be evenly distributed across the entire map area (see the standard example below). Again, make sure your GCPs are at least 50 ft from the boundary of your map.”
Maybe I’m too old school, but, if 10 GCP’s make for adequate accuracy(?), that 20 strategically placed GCP’s would still make the project more certifiable …
Maybe I’m too old school, but, if 10 GCP’s make for adequate accuracy(?), that 20 strategically placed GCP’s would still make the project more certifiable …
I agree, but it’s important to distinguish between GCPs and check points.
GCPs are used to control and align the photo block to the real world (project coordinate system). With proper distribution, I agree with the notion that 6-10 GCPs should be enough to align a properly flown project.
But I would also add check points throughout the project as well, usually equal to the number of GCPs, that are surveyed to the same accuracy as the GCPs, and are used for independent checks after the block adjustment has been done.
So, rather than 20 GCPs I would set 10 GCPs and 10 check points to be analyzed after the adjustment. IMO a project without check points is an improperly performed project with a questionable product.
“…people will come to love their oppression, to adore the technologies that undo their capacities to think.” -Neil Postman“DroneDeploy requires a minimum of 4 GCPs and we suggest using 8 GCPs for maps up to 50 acres
Seems like they’re trying to essentially tie down the corners of a assumed LDP You create with those being the limits of the AOI or project area.
a higher number of GCP doesnt do anything for better products if they’re placed poorly and irrelevantly. All comes down to deliverables and mission planning in the end.
I’m coming from the larger regional fixed wing process, so my perspective is skewed and I also got out of the arena because the smaller more flexible drone process spelled doom for my longer term survival until I actually got back into the survey pipeline.
@rover83 Finally i was wondering if someone could articulate the difference between GCP and check points. I see way to many users setting adequate GCP to adjust and align to a site but have no way to check. Great write up.
GCPs are located to a high degree of accuracy relative to the themselves and often the SPCS or local datum. Checks are not inaccurate but tend to be a slightly coarser location of various points throughout the site. GCPs coordinate values are used to attach the drone data to an defined datum. Check points are compared to post processed surface or point cloud.
I send a guy with Trimble R12 on NC’s VRS (NC provides corrections for USA, Russia, China, and Europe) and a can of orange paint. I have them spray a dot and locate its center with a 30sec shot in areas as far away from GCPs as possible. I also have them take shots in trouble areas like narrow ditches, tall grass, transitioning ground cover, areas with low reflectance variability, and even sunny vs shady areas. In briars we just take the shot with the understanding that it is of limited value.
The results from an increased amount of check shots is that I now delineate more areas as obstructed, and I have a healthy appreciation of the limits of photogrammetry. I’ve found that a lack of ground truthing is rampant in my area and have seen signed and sealed drone topos containing many feet of error. Check shots are not optional regardless of LiDAR or RTK or PPK or manned aircraft flights. We’re in the business of proving accuracy, not just regurgitating factory specs.
I recently reviewed some processed data from a site where a PLS had provided bad GCP data. Relative to itself, the data was +- 2ft horizontal. Vertical was within a tenth except for one GCP where it was off 28ft. I was surprised to see that the Pix4D processing results squeezed the 28ft of error into an average of 4ft per eight GCPs in a 300 acre site. I would have thought the software would have detected a single GCP with a radically bad elevation. Does anyone have any technical papers explaining how the Pix4D or other software use the GCP values to increase (or decrease) error?
I don’t have anything in front of me right at the moment, I’ll see if we have anything I can post up.
I don’t use Pix4D very much any more, but I seem to remember a lot of H/V errors being tied to either failing to use the targets as GCPs (marking them as checkpoints only) and/or failing to run a camera calibration.
“…people will come to love their oppression, to adore the technologies that undo their capacities to think.” -Neil PostmanGCPs are required even with RTK. Maybe less but still required.
GCPs are required even with RTK. Maybe less but still required.
With our high-end unit we run a PPK+IMU loosely coupled solution for typical lower altitude flights, and don’t use the targets for anything more than check points. Well, almost. Every few projects we’ll bump the LiDAR data up or down a couple of cm to better align it with the targets, but I don’t think we’ve ever messed with the horizontal.
But that’s with a high quality inertial system and running a local base to process against.
“…people will come to love their oppression, to adore the technologies that undo their capacities to think.” -Neil Postman
Log in to reply.