|Posted by Aaron Corcoran on July 31, 2015 at 1:10 AM|
I just got back from conducting field research on bats and insects at the Southwestern Research Station, my seventh summer in a row conducting research there. There are many reasons I keep going back. The station is home to one of the most diverse communities of bats and moths in North America. There are world-class field courses on both taxonimic groups, and I've been involved with both courses, one as an assitant, one as a guest lecturer. This diversity ensures I won't run out of animals to study any time soon. I also have at my disposal a small army of eager volunteers, ample lab space, a workshop where I can build anything I might need, three meals a day (lunch, dinner, and second dinner), and a community of researchers, volunteers, and staff that eases the burdens of long nights that seem to stretch one into the next.
Volunteers Theo, Julia and Alexa sort insects in the lab. My assistant Orion pretends to eat a sphinx moth
For several years now, I've been developing a field-recording setup for capturing bat-insect interactions on multiple video cameras. I can then use these videos to reconstruct the flight paths of predator and prey in 3-D. By 2011 I had refined the system to be able to capture volumes of about 125 cubic meters, or about half the size of a racketball court. I considered this quite a feat, but the system had limitations. I was able to document predatory events of bats that attack insects close to the ground, such as Myotis bats. However other species, such as Mexican free-tailed bats, fly exclusively in the open, far off the ground and outside the view of my cameras.
This summer I deployed the next version of this setup involving four sets of high-resolution infrared cameras and high-power infrared lights. With our new apparatus we can capture volumes of 1000 cubic meters or more. In other words, we can detect and reconstruct the precise 3-D position of a small moth anywhere within a volume the size of half of a basketball court and up to three times the height of the basket. Each night is a bit like setting up for a rock concert. In the event of rain (we work in the monsoon season) we race frantically to unplug and then cover all the equipment, letting it dry throroughly for at least 24 hours before plugging it back in.
Time-lapse photo of field recording setup, including three cameras and IR lights, and an ultraviolet light to attract insect activity. Traces of bats and insects can be seen in the night sky.
In order to use videos from multiple cameras to reconstruct flight paths in 3D you need to know the exact positions, orientations, and specifications of your cameras. Fortunately there is now software available that allows you to derive this information through a process called 3-D calibration. Ty Hedrick at UNC and his colleagues have done a great service to science by developing this software and making it freely available. It works better than software I had paid thousands of dollars for, which I've now completely discarded.
To calibrate my cameras in the field I use a method called "Wand Calibration", where two markers fixed at a specific distance are moved in front of the cameras. Traditionally this is done by fixing two objects to the ends of a rod, or "wand". Previously I did this with an 8-foot rod mounted to a 16-foot painter's pole. This turned into what looked like an odd form of interpretive dance, and a sweaty workout. Even more of a problem was that I was limited to calibrating volumes I could physically reach with my calibration device. To calibrate cameras for 3-D reconstruction of high-flying bats and insects, I needed a new method. Enter the drone.
Field 3-D calibration kit, including UDI 818A quadcopter, IR-tape-covered wiffle balls connected by fishing line, water-baloon catapult launcher, and baseball covered in IR tape.
By drone, I mean toy quadcopter. It's amazing what $50 and some YouTube "Mods" will do, including removing the plastic protective cover and camera and attaching an anchor to tow whiffle balls covered in IR tape (OK, I didn't find that last part on YouTube). The trickiest part was finding objects that were heavy enough to ensure a taught line, but light enough that the quadcopter could pick them up. In the end my calibrations appear to be coming out alright (see image below), so I think I found a solution that works. The water-balloon launcher lets me send the IR baseball 100 feet into the air, a necessary step to determine the axis of gravitational acceleration (i.e., what's up).
Here is the output of the 3-D calibration showing the track of the markers towed by the drone, the ball launch, and the positions and orientations of the cameras:
It's way to early to be publicizing results, but I'm already seeing what looks like some fascinating foraging behavior by Mexican free-tailed bats. Here's a sample:
Perhaps most exciting is what lies ahead. We now have the technology to capture behavior of wild, freely-behaving bats and insects in large volumes and in spaces untethered to the earth. Oh yeah, and we have drones.