Autonomous, Connected and AI-embedded Multi-Sensor System for Wildlife Monitoring

The rapid expansion of human activities is significantly altering wilderness areas and their biodiversity, resulting in shifts in wildlife populations and highlighting the urgent need to implement automated species tracking systems to monitor environmental health. Current devices in all categories face real limitations, particularly the lack of local information processing, resulting in a passive system that is unable to propose an appropriate response depending on the observed event. This work presents an innovative sensing system including a new architecture for the control and computation unit, combined with a multi-sensor detection mechanism using audio analysis to improve both the device autonomy and reliability. The concept was successfully implemented in a camera trap prototype and deployed in a natural area for several months, collecting many audio and video samples that were highly valuable for training the AI classification model and experimenting with the audio analysis algorithms.