About
PAWS (Protection Assistant for Wildlife Security) is an AI-driven conservation tool built by Harvard University's Teamcore research group in collaboration with the Uganda Wildlife Authority, Wildlife Conservation Society, and the World Wide Fund for Nature. The system ingests historical poaching records and geographic data about protected areas to train machine learning models that predict where illegal activity is most likely to occur. Based on these predictions, PAWS generates poaching risk maps and recommends patrol routes that enable rangers to deploy more strategically and effectively. In its first field deployment at Cambodia's Srepok Wildlife Sanctuary in December 2018, rangers following PAWS-suggested patrol routes removed over 1,000 snares in a single month — more than double the previous rate — along with 42 chainsaws, 24 motorbikes, and one truck. PAWS is being integrated with SMART, a conservation management platform used in 800+ national parks worldwide and backed by nine leading NGOs including WWF and Wildlife Conservation Society, in partnership with Microsoft AI for Earth. To support under-resourced parks with limited historical records, PAWS can augment data using remote sensing imagery. Ongoing research extends the system with robust reinforcement learning under minimax regret, enabling resilient patrol planning even under uncertainty about adversary behavior. PAWS is primarily aimed at conservation organizations, park administrators, wildlife rangers, and academic researchers working on green security problems.
Key Features
- Poaching Risk Prediction: Uses machine learning trained on historical poaching records and geographic data to predict where and when illegal activity is most likely to occur.
- Optimized Patrol Route Generation: Generates strategic patrol route recommendations for rangers based on predicted risk areas, enabling more effective deployment of limited resources.
- SMART Software Integration: Integrates with SMART, a conservation management platform used in 800+ national parks globally, enabling broad deployment of PAWS predictive models.
- Remote Sensing Data Augmentation: Incorporates satellite and remote sensing imagery to support under-resourced parks that lack extensive historical poaching records.
- Robust Patrol Planning Under Uncertainty: Applies minimax regret reinforcement learning to handle uncertainty in adversary behavior, producing more resilient patrol strategies.
Use Cases
- Conservation managers use PAWS to identify high-risk poaching zones and deploy rangers more strategically across large protected areas.
- Wildlife rangers use PAWS-generated patrol routes to maximize snare detection and removal during routine field operations.
- Park administrators integrate PAWS with SMART software to centralize conservation planning, reporting, and patrol coordination.
- Research institutions use PAWS's reinforcement learning framework to study and develop robust patrol strategies under uncertain adversary behavior.
- Under-resourced conservation organizations use PAWS with remote sensing data to launch AI-driven anti-poaching programs even with limited historical records.
Pros
- Proven Real-World Impact: Field tests in Cambodia showed rangers doubled snare removals in the first month, demonstrating clear, measurable effectiveness.
- Global Scale via SMART Integration: Integration with SMART conservation software enables deployment across 800+ protected areas worldwide with minimal adoption friction.
- Backed by Leading Conservation Organizations: Developed with WWF, Wildlife Conservation Society, and Uganda Wildlife Authority, ensuring strong domain expertise is embedded in the system.
- Supports Under-Resourced Parks: Remote sensing augmentation means even parks with limited historical data can benefit from AI-driven patrol optimization.
Cons
- Relies on Historical Poaching Data: Machine learning models are most effective with robust historical records, which may be scarce for newly established or data-poor reserves.
- Academic Research Tool: PAWS is a research project rather than a commercial product, which may limit documentation, user support, and ease of adoption for non-technical staff.
- Limited Self-Serve Access: Deployment is primarily managed through institutional partnerships, making it difficult for independent conservation groups to adopt without formal collaboration.
Frequently Asked Questions
What is PAWS and who developed it?
PAWS (Protection Assistant for Wildlife Security) is an AI-powered anti-poaching tool developed by Harvard University's Teamcore research lab in collaboration with the Uganda Wildlife Authority, Wildlife Conservation Society, and the World Wide Fund for Nature.
How does PAWS predict poaching activity?
PAWS uses machine learning algorithms trained on historical poaching records and geographic data about protected areas to model poacher behavior patterns. It then generates risk maps highlighting zones with the highest probability of illegal activity.
Has PAWS been tested in real conservation settings?
Yes. The first field test was conducted in December 2018 at Srepok Wildlife Sanctuary in Cambodia. In the first month, rangers using PAWS patrol suggestions removed over 1,000 snares — more than double the prior rate — plus 42 chainsaws, 24 motorbikes, and one truck.
How does PAWS integrate with existing conservation management tools?
PAWS is being integrated with SMART, a widely used conservation management platform deployed in 800+ protected areas globally. This integration, developed with Microsoft AI for Earth, allows PAWS predictions to plug directly into existing ranger planning workflows.
What if a park doesn't have enough historical poaching data?
PAWS can supplement limited historical records with remote sensing satellite imagery, allowing under-resourced parks to still benefit from AI-driven patrol planning even without extensive prior data.
