Much has been considered when it comes to protection of the African elephant, whose ivory makes them a high-prized target for poachers. While the species is currently vulnerable, which means its population is increasing but only incrementally, teams of rangers patrol national parks and the other protected habitats where the African elephant tends to roam. However, poachers often remain a step ahead, taking advantage of a dispersed elephant population and the limited number of rangers tasked with protecting them.
Now, those rangers are getting an assist from artificial intelligence. PAWS, which stands for Protection Assistant for Wildlife Security, is a machine learning algorithm-based technology that helps rangers predict where poachers are most likely to strike while also developing randomized patrolling paths for the rangers to follow. Developed at the University of Southern California-Los Angeles, the algorithm incorporates data from a given patrol zone and previous patrol findings about the poachers’ behavior to formulate an array of suggested future routes. The A.I. relies upon game theory and predictive behavior modeling in formulating its output results for rangers to use as guidance.
A team of USC Ph.D. students, headed by professor of computer science Milind Tambe began working on the systems that would ultimately become PAWS in 2013. The algorithms that Tambe’s team would ultimately come up with are based on security game theory and are not dissimilar from theories used by the Department of Homeland Security, the Transportation Security Administration, and the Coast Guard.
Uganda’s Queen Elizabeth National Park was the site of a preliminary field test of PAWS in April 2014. In constructing plans for how to save the threatened African elephant species or any highly-sought after target population from poachers, it’s critical to stay ahead of the poachers, locating them and their traps before they strike. With an estimated 96 African elephants killed per day, these slow-moving, graceful behemoths make for a logical testing population of PAWS technology.
As rangers observe and input sightings of traps, snares, animal remains, other signs of poaching, and even mere animal sightings, PAWS is able to more accurately predict recommended patrol routes. Since initial testing in Africa and Malaysia, the tech has been shown to outperform non-A.I. assisted patrols in terms of animal and human sightings per kilometer.
“It’s very easy for us just to do theory and mathematical models and be seduced by that,” Tambe says. “But seeing these places in person had the subtle impact of saying that we should protect this and not let it disappear off the face of the Earth.”