Digital Platform for Traffic Safety-Risk Prediction in Rural Areas
The project investigates the feasibility of merging vehicle sensor data with safety-related data to innovate traffic risk prediction.
In the D-TRAS project, a digital platform is being developed to warn road users (car and motorbike drivers) of possible traffic risks, mainly on roads in rural areas. This includes mountain roads in particular. A demonstrable traffic risk can arise if the driver is not concentrated or distracted, or if there are special environmental or road conditions, or if they change unexpectedly. Information collected and processed for the driver, or corresponding warnings of changing visibility conditions, traffic situations, road conditions, accident and wildlife hotspots, as well as recorded behaviour of other road users, should help to increase safety on roads in rural areas and reduce the number of accidents involving motorcyclists and car drivers. Here, data from networked vehicles can contribute relevant information on the current traffic situation.
In cooperation with organisations in the field of digital mobility from Austria and Germany, different data sources are bundled, analysed and evaluated. The D-TRAS platform also includes an AI model (artificial intelligence), which generates a usable number of traffic information for the road users from a multitude of information and presents this to the driver via a smartphone or a wearable device.
Researchers from Austria (Motobit GmbH and VIRTUAL VEHICLE) and Germany (University Göttingen, Caruso and NEXT) are working on achieving the research goals in the D-TRAS project. In a study with at least 100 road users, the D-TRAS platform will be evaluated in two European regions, Styria (Austria) and the Harz (Germany).
The information available through the D-TRAS project and the traffic information generated during the study will provide new opportunities for traffic planners and road maintenance managers to improve traffic infrastructure.
How the information flows
The following figure illustrates the levels of information flow. The platform collects a wide variety of data from the vehicle as well as from other existing data sources. (A) This collected data is pre-processed in the vehicle (B) and then filtered in a central, digital platform according to road safety relevance and predicted for the driver. (C) These risk warnings can then be displayed in real time via different devices (D).
AI model training
A central element for forecasting relevant traffic risks in the D-TRAS platform is the AI model, which generates warnings for motorbike and car drivers from different information and conditions.
The training of this platform is done with historical data, which means with already existing data, and also with dynamic data, which means all those that are constantly updated.
What is VIRTUAL VEHICLE’s role in this project?
VIRTUAL VEHICLE is the coordinator of the project and is responsible for the project management, the development of the D-TRAS platform architecture, the D-TRAS kit for mobile apps and the presentation of the results. In addition, a research team from the Contextual Information Systems and Management group is working on building the D-TRAS platform as a cloud solution.