Tiny monitoring devices, strapped to birds, used artificial intelligence to work smarter, it is claimed.
Computer scientists and ornithologists at Japan’s Osaka, Nagoya, and Tohoku universities even went as far as saying the gadgets formed the world’s first “first AI-enabled bio-logger” experiment.
The team designed and built the gizmos, each containing a video camera, an accelerometer, GPS unit, and a Microchip ATmega328P 8-bit microcontroller. The hardware was attached to black-tailed gulls and streaked shearwaters on the Kabushima island near Hachinohe City in northern Japan.
Crucially, the software running on the equipment was trained to turn on the camera and start recording whenever a bird was deemed to be foraging. It did this by taking readings from the accelerometer and GPS unit, and running them through a decision-tree model so that it could determine whether or not the birds were searching for food. If they were, the code would turn on the camera and start recording. This turned out to be more efficient than randomly turning on the camera.
Eagle-on-EGLE* violence: American icon sends govt-flown drone hurtling into the waters of Lake Michigan
For instance, if the model identified five consecutive bouts of flying behavior associated with hunting, the camera was automatically activated.
“The new method improved the detection of foraging behaviors in the black-tailed gulls 15-fold compared with the random sampling method,” said Joseph Korpela, lead author of a paper describing the experiment and a researcher at Osaka University. This also increased the amount of data the camera and sensors could gather between recharging tenfold, from two hours to 20, we’re told.
You can this in action in the video below.
“Since bio-loggers attached to small animals have to be small and lightweight, they have short run times and it was therefore difficult to record interesting infrequent behaviors,” said Takuya Maekawa, co-author of the paper and an associate professor at Osaka University, late last week.
The study was published in the Communications Biology journal.
The team reckoned extending bio-logging experiments with AI will help biologists study animal activities more easily in the wild by capturing specific moments of behavior that are difficult to observe without full-time monitoring.
“These systems have a huge range of possible applications including detection of poaching activity using anti-poaching tags,” Maekawa concluded. “We also anticipate that this work will be used to reveal the interactions between human society and wild animals that transmit epidemics such as [the] coronavirus.” ®