Arctic Wildlife Monitoring in Greenland Using Edge AI

The monitoring of wildlife populations in the Arctic region such as Polar bears (Ursus Maritimus) and Arctic foxes (Vulpes lagopus) presents significant challenges due to extreme weather conditions and limited surrounding infrastructure. In addition, certain animals pose a high safety risk to researchers conducting fieldwork in Greenland, therefore the detection of such animals is of the utmost importance to prevent direct human-animal encounters. This scientific work integrates a camera-based device with an edge AI (artificial intelligence) model for real-time object detection. The proposed object detection model is deployed in an embedded environment based on a single-core microcontroller, allowing efficient inference with reduced power consumption and latency compared to conventional computer-based solutions designed on higher abstraction levels. Hardware design considerations, like component selection, manufacturability are discussed in detail. The system is evaluated through comparative analysis against alternative hardware platforms and design approaches. Finally, the thesis outlines potential extensions for the proposed system, moreover highlights the broader importance of wildlife monitoring applications.