D-TRAS - Digital Platform for Traffic Safety-Risk Prediction in Rural Areas
Partner: Virtual Vehicle Research GmbH; motobit GmbH; NEXT Data Service AG; Chair of Information Security and Compliance, Georg-August-Universität-Göttingen; Caruso GmbH
Project duration: 02/2021 - 01/2024
Funding: BMWi / FFG
Methods: Machine Learning, Data Collection, Field Studies, Cloud Platform Architecture Development, AI Predictive Model Training, Business Model Design
The Austrian-German joint research project D-TRAS investigates the feasibility of combining vehicle sensor data with safety-relevant data to innovate the prediction of traffic risks. The aim is to provide a traffic risk prediction service for road users running on a digital platform considering the contextual information provided.
Collected dynamic sensor data from a driver’s vehicle, smartphone, or wearables will be aggregated into safety-relevant information and transmitted to a digital platform operated in the cloud where it is enriched with data from other users and data marketplaces. A trained AI model will then be used to predict spatiotemporal traffic safety risks. Thereby, road users can be warned when approaching a traffic risk to improve road safety. In addition, options for feedback and evaluation will be provided to improve prediction quality.
The D-TRAS concept will be validated in two regions with different topology and usage behavior, the Styria (Austria) and the Harz Mountains (Germany). A business model architecture matching the technical architecture will be developed in a final step considering barriers to user participation and possibilities for sustainable operation of the platform.
In the project, the Chair of Information Management leads the training and implementation of the AI model.