Yet despite this progress, safety gains are plateauing. Because recognising what is in a scene is not the same as understanding what is about to go wrong. Road safety is not driven by objects alone. It is driven by interactions, context, intent, and constraints. A pedestrian stepping off the curb, a vehicle drifting across a lane line, or a cyclist changing trajectory can be harmless in one setting and a near-miss precursor in another. Current systems, however, still treat road users as isolated entities rather than participants in a dynamic, rule-bounded environment.

This keynote argues that the next breakthrough in safety requires a shift from object-centric detection to grounded traffic scene understanding. Instead of producing only bounding boxes and trajectories, we move toward structured, evidence-based descriptions of what is happening: who is yielding to whom, which movements violate expectations, and which behavioural patterns quietly repeat before serious crashes occur. This reframing turns raw sensor data into reviewable safety events that support proactive risk intelligence, not just reactive enforcement.

Recent advances in computer vision, especially vision-language models, make this shift practical for the first time, when used as constrained semantic layers anchored to visual evidence. The keynote closes with deployment priorities: privacy by design, auditability, robustness across locations, and evaluation based on real safety outcomes rather than detection accuracy.

Short-Bio:

Prof. Dr. Ansar Yasar is a Professor at the Transportation Research Institute (IMOB) of Hasselt University in Belgium, one of Europe’s leading centers in the field of road safety, intelligent mobility, and transport management. He leads the educational and research activities in Intelligent Transport Systems (ITS) and smart mobility, and currently serves as the Head of the Business Development Unit at IMOB. In this capacity, he connects scientific research with real-world mobility applications through projects such as Safe Routes to Schools, Safe-Trucking Systems, and AI-based traffic safety platforms.

Prof. Yasar coordinates the VITRONIC Research Chair on AI-based Traffic Enforcement and Behavioural Change, focusing on intelligent systems for real-time violation detection, gamification, and sustainable driver behavior improvement. He is also the co-inventor of U.S. Patent No. US12425873B1, “Method for Managing Coverage and Interference of Internet of Drones (IoD),” which proposes advanced methods for optimizing drone and IoT communication networks.

He has played a leading role in several European research programs, including FP7 DATA SIM, ERA-NET Smart-PT, and Horizon projects such as iSCAPE and Track&Know. His broader research interests include smart cities and communities, connected and automated mobility, drone system management, data-driven road safety, and sustainable transport solutions.

Prof. Yasar holds a Ph.D. in Engineering from KU Leuven (Belgium), an M.Sc. in Computer Science and Engineering from Linköping University (Sweden), and a B.Sc. in Software Engineering from Foundation University Islamabad (Pakistan). He has authored and co-authored more than 250 peer-reviewed publications, co-edited the book “Data Science and Simulation in Transportation Research,” and regularly serves as a reviewer and expert evaluator for European innovation programs such as EUREKA, COST, and Horizon Europe.