FAIR-AID: AIoT System for Bias-Aware Search and Rescue in Mountain Environments
FAIR-AID is an innovative research project focused on developing a bias-aware AIoT system for enhancing search and rescue (SAR) missions, especially in mountainous and disaster-prone environments. The system integrates advanced sensors, open-source drones, edge computing, and TinyML to reduce response times, improve detection reliability, and address AI bias challenges in critical scenarios.
Project Objectives
- Design and implement a human-centered AIoT architecture tailored for SAR operations.
- Develop hardware modules: Body-Worn Sensing Module (BWSM), Edge-Enabled Open-Source Drone Module (ENOSDM), and Ground Monitoring Center (GMC).
- Collect and annotate multimodal datasets (IMU, acoustic, GPS, visible/multispectral/thermal images) through real-world experiments.
- Design and evaluate bias-mitigation AI algorithms using pre-processing, adversarial training, and fusion learning techniques.
- Integrate and validate the full system to ensure reduced rescue time and improved equity in victim detection.
Research Focus
- Human-centered AI for emergency response
- Bias mitigation in AI algorithms
- TinyML for low-power rescue devices
- AIoT architectures for multimodal sensing
- SAR applications in mountainous and disaster-hit environments
Consortium and Partners
The project is implemented by the National University of Science and Technology Politehnica Bucharest.
Principal Investigator: Dr. Ana-Maria DRĂGULINESCU
Implementation Timeline
Start Date: March 28, 2024
End Date: March 28, 2025
Funding
Funded by L'Oréal and UNESCO through the “For Women in Science – Young Talents Romania” program 
Total budget: €10,000
Expected Outcomes
- List of SAR use cases, system requirements, and modular architecture (BWSM, ENOSDM, GMC)
- Annotated multimodal datasets for AI training and bias analysis
- AI algorithms that mitigate gender, age, and race biases in victim detection
- Integrated FAIR-AID system validated in real-world SAR scenarios
- Project web platform, scientific publications, workshop, and patent analysis
The FAIR-AID project aims to redefine search and rescue through human-centric AI and open innovation, while promoting responsible AI research and inclusivity in life-saving technologies.