From Reaction to Prediction: Re-imagining Traffic Safety Through AI and Intelligent Transport Systems
Please check the following document for the lecture slides by Prof. Quddus.
SPEAKER Prof. Quddus
Chair Professor of Intelligent Transport Systems (ITS)
Department of Civil and Environmental Engineering
Imperial College London DATE AND TIME 02 SEP 2025, 7 pm -8 pm
Venue
Rayson Huang Theatre, The University of Hong Kong ORGANISED BY Institute of Transport Studies, The University of Hong Kong
REGISTRATION LINK
ABSTRACT This talk presents a paradigm shift in traffic safety analysis, moving from reactive approaches based on historical collision data to proactive conflict detection and predictive collision modelling. In the reactive phase, advanced statistical methods are used to analyse collision patterns, identify high-risk locations and evaluate the effectiveness of safety interventions, laying the groundwork for evidence-based traffic safety evaluation programmes. The proactive phase leverages vehicle-based conflict data to anticipate potential collisions before they occur. To improve interpretability and accuracy, a context-aware conflict prediction algorithm is introduced, combining traffic covariates with vehicle sensor data using a hierarchical Bayesian threshold-excess model grounded in Extreme Value Theory (EVT). This approach improves model performance and enables efficient real-time deployment. In the predictive phase, vision-based video analytics are used to detect and explain pre-crash behaviours, informing the development of more transparent and trustworthy Advanced Driver Assistance Systems (ADAS). Together, these phases represent a fundamental shift—from reacting to crashes after they occur, to preventing them through intelligent, explainable and data-driven systems. BIOS Professor Quddus is Chair Professor of Intelligent Transport Systems (ITS) in the Department of Civil and Environmental Engineering at Imperial College London. He is internationally recognised for his pioneering research in transport safety and simulation, autonomous and connected vehicles, big data analytics, and map-matching technologies. His seminal work on AI-based map-matching has become highly influential, widely cited by researchers, and adopted by the ITS industry, automotive manufacturers, and National Highways. Notably, his award-winning risk-mapping algorithms have been implemented to inform strategic safety policies and procedures across the UK’s 4,300-mile strategic road network. Over the past two decades, Professor Quddus has led and co-led research projects totalling more than £12.2 million, funded by UKRI, the European Commission, government bodies, and industry partners. He has directed large-scale research programmes, including a mini-CDT, and served on international research committees such as the Transportation Research Board (Washington, DC), while also contributing as an editor of leading peer-reviewed journals, including Transportation Research Part C: Emerging Technologies. To date, he has supervised over 40 doctoral and postdoctoral researchers and authored more than 140 peer-reviewed journal publications.
