PAWS(Protection Assistant for Wildlife Security )

PAWS advances research in machine learning, AI planning, and behavior modeling for assisting in protection of wildlife. PAWS takes basic information about the protected area and information about previous patrolling and poaching activities as input, and generates predictions of potential poaching locations and possible patrol routes as output. The core algorithm of PAWS integrates machine learning for predicting poachers' behavior, game-theoretic reasoning and route planning. More specifically, PAWS learns the behavior models of the poachers from the crime data collected. Based on the poachers' behavior model, PAWS calculates a randomized patrolling strategy, in the form of a set of patrol routes and the probabilities of taking each route. PAWS then suggests patrol routes sampled from this strategy to the patrollers.

PAWS Video (Winner of Best Application of Artificial Intelligence, AAAI Video Competition 2016)

Motivation

Poaching of endangered species is reaching critical levels as the populations of these species plummet to unsustainable numbers. The global tiger population, for example, has dropped over 95% from the start of the 1900s and has resulted in three out of nine species extinctions. Depending on the area and animals poached, motivations for poaching range from profit to sustenance, with the former being more common when profitable species such as tigers, elephants, and rhinos are the targets.

To counter poaching efforts and to rebuild the species' populations, countries have set up protected wildlife reserves and conservation agencies tasked with defending these large reserves. Because of the size of the reserves and the common lack of law enforcement resources, conservation agencies are at a significant disadvantage when it comes to deterring and capturing poachers. Agencies use patrolling as a primary method of securing the park. Due to their limited resources, however, patrol managers must carefully create patrols that account for many different variables (e.g., limited patrol units to send out, multiple locations that poachers can attack at varying distances to the outpost).

What PAWS is all About?

PAWS takes basic information about the protected area and information about previous patrolling and poaching activities as input, and generates patrol routes as output. As the patrollers execute the patrol routes, more poaching data will be collected, and feed back to PAWS. The core algorithm of PAWS integrates learning poachers' behavior model, game-theoretic reasoning and route planning. More specifically, PAWS learns the behavior models of the poachers from the crime data collected. Based on the poachers' behavior model, PAWS calculates a randomized patrolling strategy, in the form of a set of patrol routes and the probabilities of taking each route. PAWS then suggests patrol routes sampled from this strategy to the patrollers.



Preliminary Field Test

A preliminary field test of PAWS was conducted in Uganda's Queen Elizabeth National Park (QENP) in April 2014. PAWS patrols were outputted onto a GPS unit as a series of waypoints. Using the set of waypoints on the GPS as a directional guide, wildlife rangers executed their patrol and searched for signs of illegal activity. The photos below were taking during the preliminary field tests.

Predictive Analytics

Predicting where poachers will strike next is vital to protecting endangered species. By leveraging knowledge about where and when poacher attacks have occurred, Machine Learning techniques can predict where the next attack will happen. Ensembles of decision trees have demonstrated their superiority in predictive performance in both laboratory experiments and real-world field tests. Moreover, decision trees are a "white-box" approach, meaning that domain experts (e.g., conservationists, park rangers) can easily look at the learned model (in the form of logical rules) and determine whether the decision tree is making reasonable inferences about how poachers behave. Future work will focus on augmenting the patrol planning capabilities of PAWS with this new predictive analytic approach, resulting in efficient patrol schedules that are more effectively targeted to where poachers will be attacking.


How they are being poached:Yes all these snares in one year-Snares that trap these wonderful animals

USC team and collaborators patrol in a tropical forest in southeast asia.

Publications

To be published

Fei Fang, Thanh H. Nguyen, Rob Pickles, Wai Y. Lam, Gopalasamy R. Clements, Bo An, Amandeep Singh, Brian C. Schwedock, Milind Tambe, Andrew Lemieux. PAWS – A Deployed Game-Theoretic Application to Combat Poaching.To appear in AI Magazine.

2016

Thanh H. Nguyen, Arunesh Sinha, Shahrzad Gholami, Andrew J. Plumptre, Lucas Joppa, Milind Tambe, Margaret Driciru, Fred Wanyama, Aggrey Rwetsiba, Rob Critchlow, Colin Beale.CAPTURE: A New Predictive Anti-Poaching Tool for Wildlife Protection. In Proceedings of the 15th International Conference on Autonomous Agents and Multiagent Systems (AAMAS).

2016

Fei Fang, Thanh H. Nguyen, Rob Pickles, Wai Y. Lam, Gopalasamy R. Clements, Bo An, Amandeep Singh, Milind Tambe, Andrew Lemieux.Deploying PAWS: Field Optimization of the Protection Assistant for Wildlife Security. In Proceedings of the Innovative Applications of Artificial Intelligence, January 2016 (Winner of Deployed Application Award).

2016

Thanh H. Nguyen, Francesco M. Delle Fave, Debarun Kar, Aravind S. Lakshminarayanan, Amulya Yadav, Milind Tambe, Noa Agmon, Andrew J. Plumptre, Margaret Driciru, Fred Wanyama, Aggrey Rwetsiba.Making the most of Our Regrets: Regret-based Solutions to Handle Payoff Uncertainty and Elicitation in Green Security Games. In Proceedings of the 6th Conference on Decision and Game Theory for Security (GameSec), November 2015.

2015

Fei Fang, Peter Stone, Milind Tambe.When Security Games Go Green: Designing Defender Strategies to Prevent Poaching and Illegal Fishing. In Proceedings of the International Joint Conference on Artificial Intelligence (IJCAI), July 2015 (Computational Sustainability Track, Outstanding Paper Award).

2015

Debarun Kar, Fei Fang, Francesco Maria Delle Fave, Nicole Sintov, Milind Tambe."A Game of Thrones": When Human Behavior Models Compete in Repeated Stackelberg Security Games. In Proceedings of the Fourteenth International Conference on Autonomous Agents and Multiagent Systems (AAMAS), May 2015.

2014

Rong Yang, Benjamin Ford, Milind Tambe, Andrew Lemieux.Adaptive Resource Allocation for Wildlife Protection against Illegal Poachers. In Proceedings of the Thirteenth International Conference on Autonomous Agents and Multiagent Systems (AAMAS), May, 2014.

2014

Benjamin Ford, Debarun Kar, Francesco M. Delle Fave, Rong Yang, Milind Tambe.PAWS: Adaptive Game-theoretic Patrolling for Wildlife Protection (Demonstration).In Proceedings of the Thirteenth International Conference on Autonomous Agents and Multiagent Systems (AAMAS), May, 2014.

Other Publications and Workshop Papers

2016

Fei Fang, Thanh H. Nguyen, Rob Pickles, Wai Y. Lam, Gopalasamy R. Clements, Bo An, Milind Tambe.Deploying PAWS to Combat Poaching: Game-theoretic Patrolling in Areas with Complex Terrain. In Proceedings of the AAAI Conference on Artificial Intelligence (AAAI), January 2016.

2015

Fei Fang, Thanh H. Nguyen, Benjamin Ford, Nicole Sintov, Milind Tambe. Introduction to Green Security Games (extended abstract). In Workshop on Cognitive Knowledge Acquisition and Applications held at International Joint Conferences on Artificial Intelligence (IJCAI), July 2015.

2015

Fei Fang, Thanh H. Nguyen, Bo An, Milind Tambe, Rob Pickles, Wai Y. Lam, Gopalasamy R. Clements. Towards Addressing Challenges in Green Security Games in the Wild (extended abstract). In Workshop of Behavioral, Economic and Computational Intelligence for Security (BECIS) held at International Joint Conferences on Artificial Intelligence (IJCAI), July 2015.

2015

Fei Fang, Peter Stone, Milind Tambe. Planning Defender Strategies Against Attackers In Domains Involving Frequent Adversary Interaction. In Proceedings of the Fourteenth International Conference on Autonomous Agents and Multiagent Systems (AAMAS), May 2015. Extended Abstract.

2015

Debarun Kar, Fei Fang, Francesco Delle Fave, Nicole Sintov, Arunesh Sinha, Aram Galstyan, Bo An, Milind Tambe. Learning Bounded Rationality Models of the Adversary in Repeated Stackelberg Security Games. In ALA Adaptive and Learning Agents Workshop held at International Conference on Autonomous Agents and Multiagent Systems (AAMAS), May 2015.

2015

Debarun Kar, Fei Fang, Francesco Delle Fave, Nicole Sintov, Milind Tambe. Conducting Longitudinal Experiments with Behavioral Models in Repeated Stackelberg Security Games on Amazon Mechanical Turk. In Fourth International Workshop on Human-Agent Interaction Design and Models (HAIDM) held at International Conference on Autonomous Agents and Multiagent Systems (AAMAS), May 2015.

2015

Fei Fang, Thanh Hong Nguyen, Rob Pickles, Lam Wai Yee, Milind Tambe. Challenges of Green Security Games in the Wild. In International Workshop on Issues with Deployment of Emerging Agent-based Systems (IDEAS) held at International Conference on Autonomous Agents and Multiagent Systems (AAMAS), May 2015.

2015

Fei Fang, Peter Stone, Milind Tambe. Designing Defender Strategies Against Frequent Adversary Interaction. In International Workshop on Optimisation in Multi-Agent Systems (OPTMAS) held at International Conference on Autonomous Agents and Multiagent Systems (AAMAS), May 2015.

2015

Debarun Kar, Fei Fang, Francesco Delle Fave, Nicole Sintov, Milind Tambe, Arlette van Wissen. Effectiveness of Probability Perception Modeling and Defender Strategy Generation Algorithms in Repeated Stackelberg Games: An Initial Report. In Computational Sustainability Workshop held at The Twenty-Ninth AAAI Conference on Artificial Intelligence (AAAI-15), January 2015.

Background Papers

2015

Our current work on PAWS utilizes the SUQR behavioral model detailed in the following paper by Thanh Ngyuen, et al.

Acknowledgements

We thank IBM for PhD fellowship for Rong Yang.