The Role of Artificial Intelligence in Wildlife Conservation
The African forest elephant population has dropped 65% percent over the past decade as a result of the ivory trade. The rhinoceros population has fared even worse. At the end of 2015, conservationists’ best estimates indicated that there were just 30,000 rhinos left in the wild.Large scale poaching is to blame for the decimation of both populations. In fact, if current levels of poaching continue, both elephants and rhinoceroses will become extinct within the next 10 years.
However, wildlife conservationists do have a new tool in their arsenal — artificial intelligence (AI). Conservationists are using AI, along with other new techniques, to gather the information they need to predict the behavior of the wildlife they seek to protect as well as to track poachers and anticipate their next moves.
The University of Southern California Center for Artificial Intelligence in Society has begun a study that flies unmanned aerial vehicles (UAVs) around the forests to spot poachers before they strike while simultaneously locating animals. The information gathered by the UAVs is coupled with machine learning techniques to yield predictive analytics that can help anticipate poacher activity before it happens and forecast where wildlife is likely to move. They’re also using neural networks (CNN) and other techniques as part of this research.
The USC researchers, in collaboration with the National Science Foundation and the Army Research Office, have developed a new artificial intelligence (AI)-based application that uses game theoryto efficiently map patrol routes for wildlife rangers and determine what territories they should focus on.
Game theory works by using mathematical and computer models of conflict and cooperation between rational decision-makers to predict the behavior of adversaries and plan optimal approaches for containment.
The application, which is called “Protection Assistant for Wildlife Security” or PAWS, uses mathematical models to analyze data from previous patrols and evidence of poaching. The more data the researchers feed into the application, the more it can “learn” the topography, terrain, natural paths, foot traffic and animal traffic of the area that needs to be protected.
As a result, the application can map out patrol routes that make the most effective use of the rangers’ time and availability as well as the resources available to them. The application can even randomize their patrol routes, so poachers can’t pick out a pattern and adjust their behavior.
For the past two years, two non-governmental organizations, Panthera and Rimbat, have used PAWS to protect forests in Malaysia. The researchers are in discussions with wildlife authorities in Uganda to deploy the system across other sites as well.