A 1.315Mb PDF of this article as it appeared in the magazine complete with images is available by clicking HERE
With all the recent (and constant) flood of news about high-tech UAVs, it might be nice to take a break and talk about something a little less sophisticated that could also have a significant impact on aerial survey. We started our adventure into the world of photogrammetry a few years ago and quickly found ourselves testing a variety of camera systems, platforms, and software. I was amazed by the versatility and scalability of the technology. After learning how it somewhat paralleled our workflows with laser scanning, I knew we needed to invest some time in learning the ins and outs and finding solutions that would complement our current laser scanning services.
It wasn’t long after a few small scale tests at ground level that our sights drifted to the skies. Initially we tried to gain altitude by simply using a 20′ extension pole, but mounting a large, heavy DSLR camera on a long wobbly pole proved more entertaining than useful. I purchased an AR.Drone, which at the time was one of the more widely available pre-built quadrotors. After numerous random and hilarious mid-air collisions with buildings and trees, I came to the conclusion that I might not be the best pilot and that drones can and will crash. Sure, with a higher quality platform we could achieve more reliable flight, but the battle with gravity will always be present.
One night while researching aerial photography, I stumbled upon http:// publiclaboratory.org where one can purchase a simple and inexpensive balloon mapping kit. This kit–which includes a balloon, some Dacron line, and hardware–had everything we need to get off the ground minus a camera and some helium. In our first attempts we used a stock GoPro Hero 3 tied to a simple pendulum configuration as suggested by the kit, but due to the light weight of the GoPro it ended up whipping around wildly and producing too many unusable images.
After revisiting the pendulum mount may times we went back to the drawing board. With a little online research we found countless recommendations for using a Picavet suspension or a servodriven gimbal system that was either remote controlled or automatically stabilized. Going against all recommendations and logic we instead decided to keep it simple and just tied the camera directly to the base of the balloon, which in previous flights was observed to be fairly stable. Finally, some success!
At an altitude of 300 feet and a time-lapse interval of five seconds, the majority of the images were sharp, in focus, and pointed generally in the correct direction. The only defect was the 100lb Dacron tether that often made an appearance in the shot. With a slight adjustment to the mounting angle, the string was out of the picture. We soon found our rig was capturing oblique imagery and, combined with the balloon’s natural spin, it was producing excellent imagery with much more visibility of vertical structures. In this configuration the balloon’s beneficial spin is an example where being out of control is sometimes more practical than trying to fit the conditions.
The next iteration involved modifying the GoPro by replacing the stock 170 degree lens with a 90 degree lens. While the 170 degree field of view was a little wide, requiring unwarping and cropping after the fact, the new field of view was too narrow and introduced too much motion blur at a reasonable elevation. Modifying the camera, yet again, we found a 109 degree lens to be the happy medium, allowing the optimal flight altitude to settle in around 300-400 feet and producing the best results in PhotoScan.
After many more flights including survey-grade GPS ground targets we were not seeing much improvement in the resulting data quality. Noise in elevation was still close to one foot which was not acceptable for most of our applications. More brainstorming resulted in the decision to go bigger. We ordered a few larger balloons stepping up from the 3′ diameter to 6′ and 8′ and coupled them with a much better camera that was capable of low noise, high iso, and high shutter speeds. The Sony a6000 was our choice, accompanied by the Sony aspherical 16mm pancake lens. With the time-lapse application installed from Sony PlayMemories app store, we were ready to try again. The lift capacity of the larger balloon was substantially more than we had estimated. The new heavier camera provided no challenge in terms of lift capacity and by adding additional weights we determined that our ~5-6′ diameter balloon was capable of lifting five pounds while our Sony camera weighed less than one pound.
The excess lift capacity made the line so tight and vertical that instead of walking with the balloon to keep it from losing altitude due to wind resistance we could instead jog at a steady pace and the balloon could keep up. The flight time for this example was about 30-40 minutes and after a night of processing in PhotoScan the results were much more impressive.
Not bad for less than a $1500 hardware investment, right? I would surmise that similar results can be achieved with a cheaper point-and-shoot camera with a fixed lens and a manual shooting mode. Even some of the newest mobile phone cameras would potentially be up to the challenge. In reality the camera is not really much of a problem, but finding a viable source of helium is quite difficult if you do not have access to the right channels. Due to the high cost of Helium, any attempt to minimize the weight of the camera would be well worth it. If the camera is small enough one might be able to obtain a cluster of small balloons from a party supply warehouse, which is an interesting thought if ample travel is a part of your normal scanning services.
It is not my intention to deter the use of UAVs but only to provide some evidence that, while balloons might not be as fancy or cool, they work and have some traits that are worth paying attention to.
Other considerations–UAVs are fairly simple to travel with and can even be transported on most flights provided the batteries can be checked. The tank necessary to fill a large balloon cannot be loaded on planes, but helium is available at most party stores and at some industrial facilities. UAV batteries have a small environmental cost, but helium is a non-renewable resource that is growing in scarcity.
Useful Links:
chem.hawaii.edu/uham/lift.html
agisoft.com
vterrain.org/Imagery/self.html
Advanced Measurements and Modeling for Forte & Tablada Engineering Consultants of Baton Rouge, LA., specializing in cuttingedge methods and outside-the-box solutions for a wide range of disciplines.
A 1.315Mb PDF of this article as it appeared in the magazine complete with images is available by clicking HERE