PAWS anti-poaching AI predicts where illegal hunters will show up next

The illegal animal trade is a global but lucrative scourge, worth $ 8 billion to $ 10 billion annually, according to – only behind the trafficking of human beings, drugs and weapons. With so much money to be made, conservationists and wildlife rangers face overwhelming odds against well-organized poaching operations, fueled by a relentless demand for illicit animal products. The results of this protracted conflict have been nothing short of devastating for the species caught in the middle.

At the start of the 20th century, it is estimated that more than 100,000 tigers roamed across Southeast Asia. Today, due to a combination of habitat loss and aggressive poaching, there are currently fewer than 4,000 left in the wild. On the black market, products made from a single tiger can fetch as much as $ 50,000. The rhino populations were similarly wiped out, dropping from around 500,000 individuals at the turn of the 20th century to just 50,000 today. Overall, that populations of mammals, birds, fish and reptiles have declined by 60 percent since 1970.

“Poaching is the illegal hunting, capture or killing of wildlife and this is done for a number of reasons,” Erwin Gianchandani, senior adviser in the office of the director of the National Science Foundation, said on Tuesday. of a roundtable at SXSW 2021.. “Some people poach because they want to be able to claim the land the animals reside on for human use. In other cases, people poach because they want to be able to collect and use rare animal products, things like ivory or fur, even organs and skin. They often do this because they believe that these products may have religious, medicinal, nutritional or financial value. “

“It’s not only that poachers prey on the animals,” he added, “but they are often so motivated that they will end up harming or even killing the wildlife guards to escape the detection or capture of their poaching. ” Just in January, in the line of duty while patrolling Congo’s Virunga National Park, home to one-third of the world’s mountain gorilla population.

Although wildlife rangers are often outnumbered and strained in their attempts to patrol large swathes of nature reserves, AI and machine learning systems are poised to dramatically improve the effectiveness of rangers by helping them not only to track where poachers have been, but also to predict where they are most. likely to occur.

PAWS (Protection Assistant for Wildlife Safety) is one such system. Professor Milind Tambe, co-founder of the USC Center for Artificial Intelligence in Society (CAIS) and director of the Center for Research on Computation & Society at Harvard University, oversaw its development after attending a Global Tiger Initiative conference in 2013.

“I discovered how difficult things were for the animals that I had read bedtime stories to my children,” he said. .

PAWS leverages poaching data from the open-source Spatial Monitoring and Reporting Tool (SMART) system developed by the World Wildlife Foundation and uses safety games – a subset of game theory where the player must optimize limited resources to mitigate threats and attacks – to suggest the most effective routes for rangers given this historical data.

Tambe and his team first tested PAWS in 2014 at Queen Elizabeth National Park in Uganda. The park is home to a variety of endangered species, as well as thousands of traps and snares set by poachers. In addition, the 2,000 square kilometer park has only a hundred guards to patrol it. PAWS works by first dividing the park area into individual 1 km squares, then assigning a risk factor to each square based on where snares had been found before – a decade of this data collected via SMART. It then suggests patrol routes through the most at-risk areas. These suggestions change over time as poachers adjust to the actions of the guards. The time of year; locations of trails, rivers and roads; Weather and topographic conditions also play an important role in estimating PAWS. During a six-month trial period, QNEP rangers were blindly assigned a mix of patrol routes through the high- and low-risk areas of the park.

“What we found was that PAWS made higher risk predictions, in fact more traps were found,” Tambe said. “Where the paws made predictions [for lower risk] the rangers found fewer snares. “

But like all machine learning systems, PAWS is constrained by the quality of the data it ingests. “The data that park rangers collect is not perfect and there is some uncertainty with the data,” Shahrzad Gholami, a data scientist at Microsoft, told the SXSW panel. “So in the places they visit, they may not find any activity, any sign of poaching, but that doesn’t mean that poaching activity didn’t exist. Maybe it’s because the snares were well hidden. Even when rangers find a snare drum, they can only glean a lot of information from it. They can’t, for example, know if the trap was set recently or if it had been left undisturbed for weeks or even months before it was discovered. Additionally, PAWS can only address the specific act of poaching, not a poacher’s underlying motives for doing so.

Building on their success at QNEP, the Tambe team partnered with WWF in 2018 to bring PAWS to conservation areas managed by the wildlife organization, such as the Sepak Wildlife Sanctuary in Cambodia. Located along the country’s eastern border with Vietnam, Sepak is home to a large population of Asian elephants as well as bongos, antelopes, deer, macaques and leopards. Tigers also roamed the area, although none have been seen since 2007 and are believed to be locally extinct. WWF plans to reintroduce the species from 2022 and has identified Sepak Sanctuary as an ideal site to do so. But first, they must bring the poaching activity in the area under control which threatens both the tigers themselves and their prey.

Like QNEP, Sepak covers a large area, around 1,400 square kilometers, but only has 72 guards to patrol. The Cambodian Wildlife Refuge also offered a number of unique challenges in training PAWS IA not encountered in the Uganda test – such as monsoons. The PAWS team worked closely with ecologists at Sepak to develop an effective model and this collaboration led to some surprising discoveries.

“For example, it helped us discover this besides just modeling the distance of roads,” Gholami said. “We should actually specifically model the distance of a particular road called route 76, which was a main highway through the park.” The team also found that poaching practices varied depending on the country of origin of the poacher. In other words, poachers crossing the border from Vietnam behaved and reacted differently from local Cambodian poachers. The time of year was also found to be an important factor, as poachers would significantly alter the placement and distribution of their snares during the monsoon season compared to the dry months.

However, the PAWS system has proven to be very effective. “They found five times as many snares in the month the field test was underway compared to any other month on average in 2018,” Gholami said.

These are promising improvements, but conservationists still face an uphill battle against poaching. “Conservation biologists have estimated that rangers are only effective in removing about 10 percent of all traps in these parks,” said Lily Xu, a Harvard doctoral student involved in the PAWS project. “One of the most effective mechanisms to prevent poaching and other conservation crimes is deterrence, so when patrolling guards patrol certain areas, they deter poachers from returning. However, poachers expelled from an area of ​​a nature reserve often simply move their operations to a neighboring area, thus putting wildlife at risk.

Despite the challenges, Tambe’s team remains fearless. Through partnerships with WWF and other conservation organizations, Tambe hopes to implement PAWS in up to 600 protected areas around the world and expand its reach to protect marine sanctuaries and forests in the near future. to come up.

“This raises new kinds of challenges that may not arise in other areas where AI is active,” Tambe told the . “The lessons we learned would be useful for many applications; they would not be limited to wildlife crime. There are all kinds of challenges in applying AI to society and social good, and the benefits would spill over into other areas. “

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *