Aerial Art

I had the opportunity to attend ISKME’s Big Ideas Fest 2012 earlier in the week, at the lovely Ritz-Carlton Hotel in Half Moon Bay, California. On the third day of the conference, artist John Q of Spectral Q posed conference participants (along with a few hotel employees and passersby to fill in the gaps) in the shape of the golden spiral and the words “open ed.”  With everyone in position, two operators, one controlling the aircraft and one controlling the camera gimbal, used this burly hexcopter:

Untitled

to take this marvelous image:

openED Group Photo for Big Ideas Fest #bif2012
(Photo by AeriCam / Spectral Q)

The hex was truly a beast, and bested the strong onshore breezes seemingly with ease.  According to ISKME, a video will follow shortly.  Another use for the quadcopter?

Georeferencing, Fisheye Correction and Aerial Targets

Having captured a few good aerial images, I turned them over to Jason Pittman, Professor of Geosciences and project co-conspirator. He georeferenced one of those images using ESRI ArcMap 10.1 – screen grab and details below.

Here is a photo (captured last week by the copter) that has been georeferenced to a Google Earth Image of the same area.  Obvious advantage to our remote sensing data is the higher resolution imagery.  Photo is kind of wonky since I referenced it using optical cues to determine matching points in each image.  The coarse resolution of the GE image makes it difficult to line up with much precision.

The Google Earth image was georeferenced using known geographic coordinates (latitude and longitude) for the reference locations instead of relying on visual interpretation. This is a little faster and more precise.  Once we get the aerial targets we can GPS in those locations (using the more accurate Trimble GPS) and we will have known points with geographic coordinates.  Note the distortion on the north side of the image.  This is an uncorrected version of the photo and the “fisheye” effect is apparent.  The corrected mosaic you made is rad and will be a valuable step to add to our processing.  Cool thing now is that since it is georeferenced we can start over-laying existing data sets (roads, campus CAD maps, cross country trail vector file etc.).

Jason mentions the GoPro’s fisheye distortion, which is rather pronounced, and I’ve been working on correcting that.  There are many tutorials on YouTube detailing different methods of accomplishing the fix, though many of them rely on expensive, proprietary software – chiefly Adobe After Effects and Photoshop.  I’m still hunting for an open source or web-based way to do this, but since I have access to Adobe’s tools, I tried a few techniques, including processing the images using After Effects and the “cc lens” filter mentioned here, which seemed to do the trick, though some data is lost out at the edges of the photo.  In the next phase of the project, we’ll be using a different camera – likely a Canon PowerShot SX230 HS with CHDK installed – which should hopefully provide images with less distortion.

Jason also mentions aerial targets.  These are typically made from stitched 4 mil PVC, and feature corner grommets used to secure them temporarily to the ground.  They’re available from a variety of mapping supply companies and feature a few different patterns:

It seems like it would be simple enough to fabricate these DIY using foam core, so I’ll give that a shot, since we can’t fly anyway given the nonstop rain of late.

Update:  Here’s the Illustrator version of the aerial targets image, in case you want to print your own.

Map Knitter – Open Source DIY Science

After some initial quadcopter sketchiness – mainly a yaw problem that I think was the result of a loose rotor – and a couple of crashes, I completed two successful flights this afternoon down in the wetlands, after which college police offered me a ride up to campus, my first time in the back of a police car.

first ever ride in the back seat of a cop car
The next step in the project involves getting the images into the GIS. Jason Pittman (Geosciences) is close to perfecting that process, and will be sharing it here in the next week or so. In the meantime, I decided to search the web for free software to accomplish the same task, and stumbled upon http://mapknitter.org, which is a sub-project of the fabulous Public Laboratory for Open Technology and Science, at http://publiclaboratory.org.

The Public Laboratory for Open Technology and Science (Public Lab) is a community which develops and applies open-source tools to environmental exploration and investigation. By democratizing inexpensive and accessible “Do-It-Yourself” techniques, Public Laboratory creates a collaborative network of practitioners who actively re-imagine the human relationship with the environment.

In a process that I’ll document in more detail soon, I was able to correct the GoPro Hero2’s fisheye distortion using Adobe After Effects. I then brought the corrected images into Map Knitter and aligned them to Google Maps using a combination of scale, rotate, and skew.  The software is intuitive, and exports in a variety of formats, including OpenLayers, TMS, OSM-style TMS, GeoTIFF and JPG.  Here’s my first take – it’s not perfect, but it’s a start.

Knitting Aerial Images
You can view the map in context at http://mapknitter.org/map/view/2012-11-27-flc-wetlands.

Comparing DIY and Google Earth Images

This is an image from Google Earth:

This is an image from more or less the same location taken this afternoon:wetlands_01Zoomed in – note the game trails:Since the start of this project I’ve wanted to see these images side-by-side.  The thing that stands out most in my mind is the vegetation.  The DIY photos were obviously taken in a different season than the Google Earth images.  This illustrates what is to me one of the most interesting dynamics of the project, the dynamic of time.  That is, because we have the ability to generate on-demand imagery, we can capture and analyze the element of change in a very granular way.

Wetlands From The Air

wetlands_01Jason Pittman (Geosciences Professor and co-conspirator) and I had the chance to fly the quad over the wetlands today, and had a couple of really good flights.  The image above is one of the better ones – note the game trails and vegetation patterns.  The next step will be to georeference this and other photos, using known reference points – roads, property lines, that little drain (below, lower left).

wetlands_02

Camera Mount using GoPro Throwaway Plate

When the GoPro Hero 2 cameras arrived, this little plastic mounting plate was glued to the top of the each box, presumably for display purposes?

Untitled
In any case, the plate makes a really solid camera mount, albeit a shaky one.  In an effort to dampen some of the vibration and steady the camera, I cut up a piece of an old Gateway mousepad and attached it between the body of the quadcopter and the re-mounted the plate.

Untitled
Here’s the finished product, with the GoPro waterproof housing attached.  Future flight videos and still images will hopefully be a little more stable now.

Untitled

Flying Again

After the bad crash – the bitter outcome of this otherwise successful flight – I rebuilt the ELEV-8, and finally had the chance to fly it this afternoon.  One of the props is acting a little janky, so I need to check that the ESCs are all programmed correctly, and determine whether or not that particular motor has a bent shaft, but am otherwise happy with the outcome.

Programming the ESCs

Turnigy Programming CardOne of the crucial tasks in getting the quadcopter flying is programming the ESCs, or Electronic Speed Controllers.  They store a variety of parameters, not all of which I entirely understand, but all of which need to be set for each ESC.  The Turnigy Programming Card pictured above makes this process easy.  Simply connect the ESC to the card, load the settings, and repeat.  I created this little job aid as a quick reference for the appropriate settings.