Artificial intelligence at the service of operators to optimize emergency call handling.
The “Le Petit Camion” project aims to leverage artificial intelligence to detect misunderstandings during emergency calls, such as a seemingly simple yet crucial question: “You’re sending a small truck, right?”
This remark proved critical in a past intervention where a standard emergency vehicle was unable to access the incident due to the narrowness of the street. By providing an automated analysis of conversations between callers and operators, the project seeks to prevent such miscommunications.
With an average of 1,000 daily calls to SDIS 31, emergency call center operators face high-stress situations where key information can be overlooked. To address this, the project integrates AI-driven tools that analyze the linguistic content of conversations, detect speech intelligibility, assess attention levels, and recognize background noise. These features support operators in assessing situations and making decisions.
The project is built on a collaboration between researchers and field operators to develop practical solutions while preserving operators’ autonomy in their roles. It embraces an ethical use of AI to enhance emergency response management, ensuring compliance with regulatory frameworks such as the General Data Protection Regulation (GDPR).
Objectives and Expected Outcomes
Linguistic coherence to ensure that the recorded information accurately reflects what is said orally. When an operator answers a call at the Emergency Call Processing Center (CTA), they navigate multiple forms of communication: everyday language with the caller, technical language when entering information into the intervention report, and professional jargon when coordinating with other services (SAMU, law enforcement).
Analysis of non-verbal elements and the environment by developing tools capable of evaluating these aspects while complying with regulations. During a call, clues beyond spoken words can be crucial, such as the caller’s emotional state, background noise, or even the operator’s fatigue and attention levels.
Operator acceptance and trust. The project must design tools that operators trust and willingly adopt, integrating their needs and reasoning processes. The goal is not to replace operators but to assist them. Therefore, the project must ensure that new technologies do not interfere with their judgment and decision-making, remain ergonomic and intuitive, and are perceived as reliable and useful.
Data, ethics, and regulatory framework. The challenge is to work with realistic data while adhering to strict regulations, relying on simulations and training sessions. Emergency call analysis presents significant challenges, such as stringent rules on data collection and usage, while few databases exist for training AI models.
Through a collaborative approach involving multiple SDIS, researchers, and industry partners, this project will develop innovative tools to enhance call analysis reliability and support operators in their mission.
By embracing open science, the project will promote the sharing of methodologies and valuable resources across emergency services.
Ultimately, it will help optimize emergency call management and strengthen operator training, directly improving response efficiency and citizen safety.
