UAS Topo Mapping Results

andrewm · 2017-07-07T14:04:00Z

I completed a topo mapping project for a new subdivision development. The site is about 30 acres in size. Here are the specs for the survey: DJI Phantom 4 Pro Flight altitude - 180 feet Double grid pattern w/ 80% forward/side overlap 1,117 total images 10 ground control points (4 ft vinyl b&w cross targets) Trimble R8s base station setup on "benchmark" manhole Trimble R2 RTK rover Collected ~2.5 hrs of static data on the base It was very windy the day of the survey as the remnants of TS Cindy were still being felt. Ended up needing three batteries to complete the mission. Had it been calm, I probably would have only needed two. I have an as-built CAD drawing provided by the subdivision owner that had manhole "benchmark" locations on it that I planned to use as the coordinates of the base. However, after processing my static data, the horizontal location of the manhole was off by about 2.5 feet. The vertical was about 0.3' off. I'm not convinced the surveyor updated the horizontal location of the manhole on their as-built. So I ended up using the OPUS horizontal and held the elevation to what was shown on the drawing. I wanted to compare results from the UAS mapping to the as-built survey elevations anyways.I use Pix4D for my photo processing, and it took about 12-13 hours to generate the ortho, DSM, DTM, and contours. I ran the software on a server with 40 cores and 32GB RAM. I would hate to see how long this would take on my laptop! The results appeared to be very good. XYZ RMS error of the GCPs was 0.016, 0.017, 0.048 ft, respectively. I've attached the Pix4D report. Also, I did no manual editing of anything in Pix4D. I setup the project, imported the GCPs, identified each one on 5 photos, and pressed go. With additional manual processing, the results would probably be even better.For an accuracy assessment, I collected 17 additional checkpoints on various surfaces (street, sidewalk, manhole, bare ground on a lot). I used my R2 RTK rover for horizontal location of the check points, then ran a closed level loop through them from the same manhole "benchmark". I brought the checkpoints into ArcMap and ran a geoprocessing tool to extract the DSM elevation at each check point. I then calculated the difference between the DSM elevation and the Check Point elevation. The results are as follows: Minimum Difference: -0.15 ft Maximum Difference: 0.23 ft Standard Deviation: 0.10 ft 12 of 17 points were <0.07 ft different: Road: +0.04' Manhole: +0.03' Lot: +0.02' Lot: +0.02' Road: +0.01' Lot: +0.00' Road: -0.01' Lot: -0.01' Road: -0.03' Lot: -0.03' Lot: -0.06' Lot: -0.07' 5 points were >0.15 ft different: Lot: +0.23' Road: +0.23' Road: +0.20' Lot: +0.15 Sidewalk: -0.15' While I'm generally very pleased with the results, I am a little surprised by the two road shots that were 0.2' off. The sidewalk shot surprised me a little as well, but there was a lot of dirt that had washed over the sidewalk in that area that may have affected the DSM.Horizontally, the results are even better. The ortho mosaic lines of perfectly with the CAD drawing from the surveyor, except for the manholes. That's why I'm positive they only updated the elevations of the manholes shown on their as-built drawing and not their locations.What do you guys think of these results? Any critique of my methodology? I did this as a demonstration / pilot project for the subdivision builder, so it wasn't a paying job. I want to make sure I know what kind of results I can expect before I offer this as a service to my clients. Thanks!! Attached files DSLD Rose Gardens - 180ft_report.pdf (1.1 MB)

Summarize Topic

Page 2 / 2 Prev

Photogrammetry, LiDAR & UAS

Last Post by andrewm 9 years ago

22 Posts

8 Users

0 Reactions

1,336 Views

RSS

Michael Detwiler

(@michael-detwiler)

Posts: 19

Member

Translate ▼

andrewm, post: 436122, member: 10888 wrote: I assumed that was the culprit. I need to see what kind of manual editing I can do with the point cloud in Pix4D.

You can do a couple things:
1. Add manual tie points along the road.
2. Capture higher image overlap or additional images over the road.
3. Change the minimum auto tie point to 4 or 5 images. The default is 3.

Posted : July 10, 2017 9:35 pm

andrewm

(@andrewm)

Posts: 268

Member

Topic starter

Translate ▼

Michael Detwiler, post: 436355, member: 12520 wrote: You can do a couple things:
1. Add manual tie points along the road.
2. Capture higher image overlap or additional images over the road.
3. Change the minimum auto tie point to 4 or 5 images. The default is 3.

I'll try changing the auto tie point to 5 images and see what happens.

Posted : July 11, 2017 2:10 pm

Page 2 / 2 Prev

26 Forums
54.5 K Topics
592.1 K Posts
285 Online
16.9 K Members

Forum Icons: Forum contains no unread posts Forum contains unread posts

Topic Icons: Not Replied Replied Active Hot Sticky Unapproved Solved Private Closed

UAS Topo Mapping Results

Please support our supporters!

Our Policies

Quick Links

Support