graphical link to the landing page of the special report "src =" http://spectrum.ieee.org/image/Mjk5NjU1Ng.jpeg "/>

<img alt= Every year, poachers kill about 27,000 African elephants – a staggering 8% of the population. If current trends continue, these beautiful animals could disappear within a decade.

 graphic link to the landing page of the special report "src =" http://spectrum.ieee.org/image/Mjk5NjU1Ng.jpeg "/> 




<p> The solution, of course, is to stop poachers before they hit, but how to do that has long confused the authorities. In protected areas such as wildlife sanctuaries, elephants and other endangered animals can roam far and wide, while rangers can patrol only a small area at any time. "It's a two-part problem," says  Milind Tambe  computer scientist at the University of Southern California, Los Angeles. "Can you predict where poaching will occur, and can you [target] patrol it so that it is unpredictable so that poachers do not know that the guards are coming?" </p>
<p> To solve both parts of the problem, Tambe and his team created an artificial intelligence system called  PAWS, which stands for Protection Assistant for Wildlife Security . An automatic learning algorithm uses past patrol data to predict where poaching is likely to occur in the future. And a model of game theory can generate random and unpredictable patrol routes. The system has been field tested in Uganda and Malaysia with good results and by 2018 its use will extend to China and Cambodia. In addition, says Tambe, the PAWS system could soon be integrated into a tracking tool called  SMART  that wildlife conservation agencies have deployed to most of the world's sites to collect and manage data from patrol.</p>
<p> In a one-month trial with the  Wildlife Conservation Society  in Uganda's Queen Elizabeth National Park, rangers were patrolling two areas that they rarely visited, but that PAWS indicated a high probability of poaching. To the surprise of rangers, they found many traps and other signs of illegal activity. A subsequent eight-month trial looked at the entire park. Again, the patrols verified the model's predictions: In high probability areas, they found about 10 times more poaching than in low probability areas. A new trial in Murchison Falls National Park (19459012) </a> is testing whether PAWS will work as well in a different place. </p>
<p>Video: A.J.Plumptre / Wildlife Conservation Society
</p>
<p>  Andrew Plumptre  Science Director for the Wildlife Conservation Society's Africa Program, collaborates with the Tambe Group on field trials in Uganda. He says that on normal patrols, rangers enter data on what they see, using a smartphone app called  Cybertracker . About once a month, this data is uploaded to SMART. "You are able to map where the patrols searched, where they found traps and elephant carcasses and all," says Plumptre. "But there is nothing proactive about this – Ranger patrols are not enough to stop poaching." He hopes that PAWS 'predictive capabilities will make these patrols as effective and efficient as possible. </p>
<p> The PAWS system was developed because of the work that Tambe and his students began doing more than a decade ago for the safety of ports, airports, and airlines. The US Coast Guard, the Transportation Security Administration and the Los Angeles Sheriff's Department have all set up AI systems developed by the Tambe Group. And he co-founded  Avata Intelligence  in Venice, California, to commercialize this research. </p>
<p> About six or seven years ago, Tambe attended a meeting of the World Bank and attended a conference on the terrible tiger situation  of which less than 4,000 survived in the wild . "I guess I had heard about such things, but I never appreciated the scale of the problem.I suddenly realized the potential of AI to help" said Tambe, and he quickly made contact with conservation groups. </p>
<p>  Fei Fang  a former student of Tambe who is now an assistant professor at Carnegie Mellon, worked on a Coast Guard system to protect the ferry from Staten Island, New York, before turning to to PAWS. . Both scenarios are similar, she notes. "There is a defender, who is the game warden of the wildlife or the Coast Guard, and there is an attacker, who is a poacher or a terrorist, and they interact with each other in a manner as you try to predict. " </p>
<p><img alt= Predictive Patrols: Computer scientist Fei Fang in Carnegie Mellon shows a tiger trap found in a wildlife preserve in northeastern China. Fang and Milind Tambe of the University of Southern California have developed a machine learning algorithm to predict where poachers are most active.

For the PAWS team, field trials have highlighted an important reality of maintaining order in wilderness: the world is not flat. When the team started working in Malaysia, says Fang, they did not take into account the densely forested mountainous terrain. “In our first model, we took a map, divided the entire area into grid cells, plotted a line on the grid and said,” Patrols, please follow this line, “she recalls. . “We had Skype calls with them, and they said,” No, no, no, it’s not going to work. “We did not understand.”

This is only when the PAWS team visited the Malaysian reserve that they got it. “We walked with the rangers, and it took us about eight hours to do a few kilometers,” says Fang. A further refinement of PAWS takes into account geographical features that are easy to walk, such as ridges, riverbeds, and old forest trails. “We built a virtual map for the conservation area and then traced the routes according to the map.” Patrollers following the new roads found “all kinds of signs of animal and human activity,” says Fang.

At the time of going to press, Fang was in the middle of a three-month field trial of PAWS in northeastern China with the World Wildlife Fund, where the most pet worrying is the Siberian tiger. Fang says that one improvement they are working on is helping the rangers make decisions on patrol. “They can see footprints and tree marks, which indicate the direction that the poachers are heading,” she says. “And they must decide, Should I hunt poachers, what is the best strategy for changing plans if they see new information?”

Tambe and Fang also collaborate with a wildlife conservation service called Air Shepherd that uses drones equipped with infrared cameras to search for poachers at night. Their AI-based video analytics system automates what is otherwise a tedious and challenging task for humans: reviewing hours and hours of black-and-white grainy sequences and alerting rangers when Illegal activity is detected.

The next step for PAWS is to make it available to other NGOs, ideally by incorporating the algorithm into existing tools, such as Cybertracker and the SMART system. “We will probably never stop poaching completely,” Plumptre says. “But we can lower it to a lower level, so people do not decline.”

The AI ​​is generally applied to the problems of modern technology, notes Tambe, but this work is different. “We use AI to save the natural world – these beautiful landscapes and animals that we hope will not go away,” he says. “These are important treasures.”



Source link

LEAVE A REPLY

Please enter your comment!
Please enter your name here