Student tech watches over wildlife.
Wildlife conservation is critical to stemming the loss of threatened and endangered animals. Since the 1970s, it’s been estimated that there has been a nearly 70 percent drop in the populations of birds, fish, reptiles, mammals and amphibians on Earth. Protecting these species doesn’t just benefit the endangered animals themselves—humans have used animals as inspiration for everything from stronger armor to faster robots, as well as inspiration for novel medical treatments addressing issues such as stroke and diabetes.
But effective conservation efforts call for careful analysis of wildlife populations and their behavior. Engineering students at UC San Diego are now bringing wildlife tracking techniques to the next level thanks to partnerships with the San Diego Zoo—testing new methods of monitoring animals in zoo habitats as well as in the wild.
Eyes in the Sky
In order to track animals tagged with a small radio collar, conservationists used to drive around massive swaths of land as they checked for radio signal emissions to try and find marked animals—a task made even more difficult when objects such as mountains could block the radio signal. Now, students in the Engineers for Exploration program at UC San Diego have developed an aerial mapping system, flying a drone in a set pattern over the target area to map the radio signal propagation and pinpoint exactly where a tagged animal may be. The method is far more efficient than traversing often mountainous terrain, meaning conservation teams can get better data and study more animals in a shorter amount of time. The student team has already deployed the system to study iguana species in the Caribbean, and they’re hoping to test recent upgrades to the system—like offline capabilities and a better visualization map—in Turks and Caicos in 2022. “Lizards are cold-blooded, so our zoo partners want to track where they go throughout the day and map out the heat in the area to see how that affects their behavior,” says Mia Lucio, a computer engineering student and team lead.
Ears in the Jungle
Motion sensor cameras have become a common tool in wildlife monitoring, allowing conservationists to catch a glimpse of animal species in their natural habitats. But cameras are often limited by their line of sight, and even the best camera can have problems capturing small, fast-moving critters such as birds. Yet another student team from Engineers for Exploration is taking a new approach, developing an acoustic species identification system that will allow scientists to monitor the numbers and types of animals in a much larger geographic range by simply placing an audio recorder in their habitat. Acoustic data from the microphone is then run through the students’ machine learning and artificial intelligence platform, which will detect the number and frequency of different species’ calls. This information can help determine the health of bird and animal populations in remote locations.
But creating an audio classification system that can differentiate between species’ sounds is a challenge. The students are training their system on four terabytes of acoustic data—roughly 1,500 hours of audio—gathered in the Peruvian Amazon by their San Diego Zoo collaborators.
Pachyderms on the Move
A key measure of elephants’ health is their activity and movement levels, in part because movement keeps their feet healthy, preventing infection and disease. A team of engineering students working in The Basement innovation center at UC San Diego are partnering with San Diego Zoo scientists to develop a way to monitor and track the movement of these endangered species in their zoo habitats. Using existing cameras in the elephants’ abodes, the students are using computer vision to generate heat maps that show where these endangered species are spending time, with brighter colors meaning a longer duration of time spent in one area. “It shows where the elephants are hanging out, and if there are spots in the enclosure they don’t use as much, the zookeepers could encourage exploration and movement to more of their surroundings,” says Seth Litman, an electrical engineering student. Using this model, they hope to provide the zoo team with a quantitative figure capturing the extent of elephant herd movement each day. Future plans for this project, called “Zoober,” include adding the ability to track each elephant’s movements individually, but distinguishing the elephants from one another in their object detection algorithm is a tricky challenge they’re still working to solve.