There's good news for scientists who study animals that are too small to carry a GPS monitor, or that spit ID tags back out through their arms. A setup using an off-the-shelf camera can precisely capture an animal's path in three dimensions—without anyone touching the animal. Emmanuel de Margerie, who studies animal behavior at the University of Rennes 1 in France, says there are several reasons to seek new animal-tracking technologies. To put a GPS or other kind of tag on an animal, you have to capture and handle it, which can be stressful for your study subject. The tags themselves can be expensive and unreliable, and sometimes get lost. And "small species cannot be tagged at all because the tag is too heavy for walking or flying normally," he says. So de Margerie thought up a way to track animals purely by sight. He and his coauthors have dubbed it "rotational stereo videography" (RSV). It works like this: Just like our own eyesight, shooting a scene from two nearby points at once—or, in stereo—lets you measure the distance to an object. (That's why closing one eye ruins your depth perception.) Scientists have tried this kind of stereo imaging in the past, de Margerie says. But their methods usually required two or more cameras on stationary bases. De Margerie put a camera on a rotating base, so that someone shooting video can keep an animal in view as it flies or walks around. A device in the base precisely tracks the angle of the camera as it moves. The camera is set on a long, T-shaped platform that looks like a crossbow being pointed at the animal. And on either arm of the T is a mirror pointing back toward a camera sensor. This allows stereo vision with only one camera. After the user takes a moving shot of an animal, the device returns two sets of information: the stereo video, and a record of the camera's angles. Software merges these sets of data and recreates the animal's exact path. For every video frame, scientists can know their target's position in 3D space. The researchers tested their RSV device by watching a magpie walking and pecking in the grass for about six minutes. They also captured a 45-second video of a swift in flight. They found it a little harder to keep the swift in the frame of the camera as it swooped around. Nevertheless, they successfully gathered data about the bird's position and speed over time. Depending on how the animal was moving, de Margerie says, they checked its position between one and 25 times per second. However, he adds, with a high-speed camera, "Nothing prevents us from extracting 100 positions per second." Since a user has to be pointing a camera at an animal to gather data, this method obviously wouldn't work for long-distance tracking. For following animal migrations, for example, GPS collars and tags are crucial. Yet GPS tracking is only accurate to about 10 meters, de Margerie says. His RSV method could locate animals to the nearest meter, as long as they're within 300 meters of the camera. When animals are closer, within 100 meters, the device can pinpoint their positions to the nearest 10 centimeters. This technology could fill a gap between long-distance GPS tracking and up-close stationary video recording, de Margerie says; most techniques at this in-between scale "are based on expensive and heavy radar equipment, and cannot extract several positions per second." He and his coauthors built their prototype for about €5,000. De Margerie thinks tracking animals this way could be useful for studying their behavior and movement over short periods of time. For example, following a butterfly as it visits flowers in a field to understand its foraging behavior. Or learning how gliding birds soar over different types of terrain. "Any visible terrestrial or aerial species moving in open space can potentially be tracked," he says. And without tags, it will mean less hassle for the animals—as long as they don't mind being trailed by the paparazzi.
Image: by Howard/Harriet Greenwood (via Flickr)
de Margerie E, Simonneau M, Caudal JP, Houdelier C, & Lumineau S (2015). 3D tracking of animals in the field, using rotational stereo videography. The Journal of experimental biology PMID: 26056245
