Hello all, I've been toying around with the MikroKopter Hexakopter as a tool for aerial mapping and photogrammetry. After a few weeks of learning the ins and outs of the system, I finally had some time on the weekend to test the system for mapping.

In short, we placed Ground Control Points (GCPs) around a man-made ponding area at roughly ten meter intervals (~40 x 30 m in total area). The GCPs were shot in with a deferentially-corrected GPS (~0.1 m horizontal and ~0.2m vertical resolution) and with a Total Data Station (>0.03m vertical and horizontal resolution). We then flew the Hexakopter at 40m above the ground surface in an east-west grid pattern with a photograph being taken straight down every 10 m or so. The camera used was an off the self Canon A540.

The photographs were processed to remove most distortion and dropped into a photogrammetry program. A 0.05 m resolution DEM and 0.02 m photomosaic was created from 22 images. It took about two hours to collect the photographs, and shoot in the GCPs with both the DGPS and TDS. Processing of the imagery was nearly flawless and took about two hours as well. This time can probably be reduced to an hour or less.

Compared to doing this same process with a kite or blimp, I am very impressed with the Mikrokopter (and multi-copters in general) as a platform for high resolution, low altitude mapping. Let me know what you think.

Over the next few weeks, I'll be testing the same process in at a couple of archaeological sites in the southwestern US. I'll share some of those results when they are complete.

Can't wait to play with the truly opensource Ardupilot!

-Mark


Views: 3844


Moderator
Comment by Gary Mortimer on June 8, 2010 at 10:34am
Well done Mark, good job
Comment by brakar on June 8, 2010 at 10:49am
Most impressive Mark!

Was this performed by using the same method/software as in your previous thread?

Moderator
Comment by Alex on June 8, 2010 at 11:48am
thats awesome, and only in 2hours? what photogrammetry software are you using to do this?
Comment by abey on June 8, 2010 at 12:50pm
yup, i'd love to know more about the post-processing workflow...
Comment by Bart on June 8, 2010 at 1:29pm
huh? awesome!!, how t you get the z-values?? is the hexakopter expensive? plz tell us more about your project.
Comment by Peter L on June 8, 2010 at 2:47pm
Very impressive ! To reduce time, you could build a magazine holding several GCPs (discs or small cones ore something) and have the multicopter first go and place them for you !
Comment by Tom Verstappen on June 9, 2010 at 1:33am
Very nice!!!
Comment by Roberto Hawkowski on June 9, 2010 at 5:35am
Which software programs have been used to stitch the photos to reconstruct the 3D terrain?
Comment by Mark Willis on June 9, 2010 at 8:08am
Thanks for all the interest and compliments. The digital terrain model and photomosaicing used in the video were done with Leica Photogrammetry Suite. Unfortunately, it is a very expensive piece of software. I'm lucky my work has a license for it. It would be great if Photosynth (free) could easily produce similar results but we're not quite there with it yet. Several of us continue to tackle the problem of no opensource (or even close to affordable) software being available.

The post processing workflow is fairly simple:
Download images from camera.
Undistort the images in MatLab.
Download telemetry data from UAV
Enter telemetry data associated with each photograph in Leica Photogrammetry Suite (LPS).
Download DGPS coordinates for Ground Control Points (GPCs) and enter into LPS. (This is mainly to check for accuracies)
In LPS, associate the GCPs coordinates with the actual targets in the photos.
Tell LPS to generate a DEM and photomosaic
Voila! Nice DEM and Imagery.

Those are the basic steps and there can be pitfalls that cause problems along the workflow. That's essentially it tho.

By the way, the resolution of the DEM and imagery in the animation could have been much higher. I just opted for that resolution to speed up my rendering software.

Here is a Google Earth KMZ file with the full resolution data:
http://70.114.149.18/~mwillis/Hexakopter_3D_Mapping.kmz (~13.2 Mb)

My bandwidth is limited and the server is a little slow so please be patient.

-Mark

T3
Comment by Rory Paul on June 9, 2010 at 8:44am
Mark

What is your estimate on the largest area that could be mapped using this method and where are the current bottle necks (flight duration, processing time on the back end)?

Comment

You need to be a member of DIY Drones to add comments!

Join DIY Drones

© 2020   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service