AI-Enabled Biologging for Wildlife Conservation

Use Case:

Deployment of AI-powered biologgers capable of detecting animal death and selectively capturing GPS positions based on specific behavioral patterns improving tracking precision battery efficiency.

Outcome:

• Successfully trained and deployed a CNN-based embedded model on Cortex M4 microcontrollers;
• Achieved 99.7% accuracy in death detection, far exceeding the original 75% goal;
• Developed High Value Fix (HVF) mechanism for GPS acquisition based on behavior;
• Created a robust 30GB+ dataset from over 570 hours of IMU recordings on 5 raptor species;
• Enabled real-time alerting for mortality detection, improving intervention times.

Ecosystem Support:

StairwAI technical mentoring (UNIBO), AI expertise from Thingenious PC (Greece), collaboration with UTE-Halcones (Spain) for data collection and access to raptor facilities.

AI Relevance:

Sigloc_AI shows how edge AI empowers niche domains like wildlife tracking through:

1. Lightweight embedded ML deployment on constrained devices;
2. AI-powered functionalities absent from legacy products;
3. Accessibility and democratization of AI via SMEs;
4. Tangible impact on animal conservation and environmental research.

Summary:

La Siesta Technologies, a Spanish SME, leveraged the StairwAI program to revolutionize biologging tools used in wildlife monitoring. Traditionally, death detection relied on imprecise accelerometer thresholds, often requiring human confirmation. Through Sigloc_AI, La Siesta developed and validated an accurate death detection mechanism using deep convolutional neural networks optimized for embedded microcontrollers. These classifiers analyze IMU data in real time to detect mortality, triggering immediate alerts for experts. Complementing this, the biologgers feature the HVF system, which uses AI to recognize relevant behavioral patterns and trigger GPS recordings only when meaningful activity is detected. This innovation allows smaller devices with extended battery life to be deployed even on birds as light as 100g, previously excluded due to size constraints.
Sigloc_AI’s capabilities were developed with expert guidance and validated on real-life datasets from live and deceased raptors, supported by partnerships with AI specialists and wildlife experts. The result is an industry-leading solution ready for commercialization in ecological consultancy, academic research, and urban environmental health monitoring. Through its advanced features, Sigloc_AI not only enhances biologging precision but also enables a deeper understanding of animal behavior, mortality causes, and ecosystem risks—exemplifying the power of democratized AI in environmental applications.