Intelligent Traffic Risk Assessment with User Monitoring (i-TRAUM)

• Applicant: Professor Dr. Constantinos Antoniou
• Subject Area: Traffic and Transport Systems, Intelligent and Automated Traffic
• Term: since 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 571155448

The main focus of i-TRAUM is to develop a more advanced and comprehensive crash risk estimation framework, that builds on the state of the art and is modular, so that it can be applicable in multiple real-world situations and datasets. At the same time, i-TRAUM envisions to constitute the next step in driver behavior-related research, as the developed framework will be further used to investigate the evolution of crash risk and assess driver individuality, in specially simulated near real time scenarios. The i-TRAUM project will benefit from a rich naturalistic driving dataset, comprising more than three million kilometers of driving data, collected across five European countries, including Germany, across four transport modes; an equivalent of over 120 thousand trips, amounting to over 300 thousand driving hours, including user, vehicle and road contextual data collected under real time driving conditions and containing representative instances of normal driving, near misses and incidents. Workpackage (WP) 1 will augment the existing dataset with historical crash occurrences and road network data. Part of this analysis is also to identify the optimal data aggregation period for the various variables that will be used. WP2 will develop the framework for the crash risk estimation. This WP will use the pre-processed data and methods developed in WP1 to consolidate all those factors into one framework, evaluating their individual significance. The first objective of WP3 is to use the developed i-TRAUM framework and data from the i-DREAMS database to further knowledge in the field of driver behavior analytics and draw valuable conclusions in two key topics: temporal dynamics and driver individuality. A second objective of WP3 is to explore the applicability of Large Language Models (LLMs) as learning tools in microscopic driving behavior studies. WP4 will leverage the outputs of the previous WPs to obtain useful insights for the design of crash risk estimation frameworks and their suitability for real-world applications. Especially the developed crash risk framework will be released in an open access format, which could be used in the design and deployment of relevant commercial Advanced Driver Assistance Systems (ADAS). The research findings will be communicated to relevant stakeholders (academia, industry, and policymakers) through journal publications, conference presentations and dedicated workshops, among others.

• DFG Programme: Research Grants