Drone-mapping for highway asset management - Vassilis Polychronos - GeoSense

Pix4D Café
1

Hi everyone,

Today, our featured project is a highway asset management mission in northern Greece:

You can navigate, and measure the project on Pix4Dmapper Cloud: https://cloud.pix4d.com/pro/project/263195/model?shareToken=ac1a042c-b241-427f-98b6-24018a951613

This is the story of this project:

“The goal of the project was to produce documentation for asset management of 11km of new highway and an as-built survey on a radius of 80 meters around it.

Our client required an accuracy of 10 cm.

The highway is in northern Greece, on a plain surrounded by mountains. The significant variations in terrain elevation posed a challenge, with the requested accuracy level in mind, we had to plan and adapt our mission to these topographic conditions.

We used a senseFly eBee Plus RTK drone with a S.O.D.A. camera. To plan the mission, we used eMotion, the planning software from senseFly that is capable of designing corridor mapping missions following SRTM DTM models.

In less than two hours, we captured 1300 images, mapping 11km of highway, covering an area of  3.7Km__2

We processed the dataset using Pix4Dmapper software on a desktop device. The RTK GNNS receiver in the eBee Plus provided solid and highly accurate geotags to the images.

The resulting product had a mean GSD of 3.17cm.

We assessed the accuracy of the project using 4 checkpoints where the mean DX and DY was 1.2 x GSD and DZ was 2 x GSD. These results were below the 10cm accuracy required by our client.

These are final deliverables we provided to our client:

  • _       _3D point cloud in las format
  • _       _3D textured mesh in .fbx format
  • _       _Orthomosaic
  • _       _DSM

The benefits of drone-mapping for highway asset management go beyond time savings -over 80% compared with traditional surveying techniques-; as it is unobtrusive,  it possible to perform the survey while the highway remains fully operational, without disrupting the normal traffic.

The rich data and highly accurate deliverables produced with Pix4Dmapper outperformed the initial request providing a complete and clear real-world digital reconstruction.” Vassilis Polychronos, CTO of GeoSense.

Technical details

  • Device used: eBee Plus RTK drone with a S.O.D.A. camera

  • Image capture path: corridor mapping

  • Number of images: 1300

  • Pix4D software used: Pix4Dmapper

  • Processing platform: desktop device

Enjoy your virtual visit, feel free to ask questions, share comments and your own projects.

Pix4D team

3 Likes
2

Great project! Could you please tell somthing about the desktop hardeware and the processing time?

3

Thank you for commenting. The hardware was a moderate PC. A 6-core i7 6850k CPU at 3.6Ghz, 32Gb with an SSD HD. Total processing time for all 3 steps was about 7 hours. Sorry about the user name “Reseller GeoSense” as it is our corporate account. I am Vassilis, feel free to ask.

2 Likes
4

That is an awesome project, Vassilis. The sheer size of the mapping is incredible. Could you share with us some of the workflow for your client “after” the processing? The client wishes to measure? Or just for recording/tracking purposes? Which tools do they plan to import the outputs/results into?

Ideally Pix4D could provide the necessary tools so that the results could just be shared with the client and they could perform all/most of their tasks directly in Pix4D Cloud or pix4Dmapper Desktop.

2 Likes
5

Thank you, Hugo. The customer uses a special software that drapes the othomosaic on the DSM and creates a 3d representation of the study area. Upon this 3D area, the software provides tools for digitization. He digitizes in vector format lines, points, and polygons. Also, he creates TIN and measures volumes.

For the scope of the project, the customer used Pix4D products in his software to digitize all the infrastructure around the highway, like light poles, rainwater grids, fuse boxes, electric and electronic equipment containers, traffic signs and all the existing human-made assets like bridges, aggregates etc.

Finally, all the vectors are stored in the project’s digital record system inside a GIS software.

3 Likes
6

   Hi, nice project. I also did a 2000 photos projects, but the processing time was much longer than yours , around 18-28 hours. Could you share some setup details for all three steps? Was it on full or rapid , point cloud optimal or high , image scale 1/2 ?

It could help me reducing the processing time. 

I have a PC : Intel(R) Xeon(R), E5-2620 v4 @ 2.10 Ghz (2 processors) CPU, 128 GB RAM , and SSD

Thank you! 

Mapit Cluj - Romania 

7

Hi Mapit Cluj,

For the processing, I followed Pix4Dmapper 3D Maps template. I believe that you have much longer processing time because of the processor you have. I own a second workstation with 2 x XEON 5620 4Core each at 2.4Ghz, 64Gb of RAM and SSD and It performs really bad compared to the humble i7. It takes about twice as much time to process the same dataset compared to the i7. It has to do with the clock speed. Also, I think that Pix4Dmapper doesn’t like old hardware and especially older XEON CPUs. 

Just for your records, the above i7 has processed 3000 images @ 20mpixel in 26h. 

Thank you,

Vassilis 

8

Hello again, the 3D textured mesh in .fbx format did you use it for autocad? Because I’ve had problem in loading it in realistic color.

9

Hello to you again. We do not use Autodesk software. We find it very hard to cope with UAV photogrammetry big data. The fbx was used for visualization purposes (sketchfab).

Thank you.

Vassilis