AI aims to predict crash risks before accidents happen
Weather data and sensors provide information that helps identify accident risks. But how are people warned about them? Researchers at Chemnitz University of Technology are finding out.
Fog on the highway, a slippery road surface after a curve, or the end of a traffic jam beyond a hilltop – anyone behind the wheel who notices such a hazard too late has little time to react. Police recorded approximately 98,000 traffic accidents in Saxony in 2025. A European research project is now aiming to detect such risks early on. Researchers from Chemnitz University of Technology are participating in the project.
The project is called ROAD4SAFE and launched in October 2026. Several universities, research institutions, and companies are collaborating on it with the authorities responsible for road maintenance. The European Union is supporting the project with approximately five million euros.
AI could help detect hazards
The project participants are relying on artificial intelligence for their solution. It is designed to collect and analyze information from weather and traffic data, past accidents, road sensors, and connected cars. This will generate predictions about where conditions will become dangerous for drivers.
But that’s not all. A computer system is designed to help avoid these hazardous situations and take corrective action in time to prevent accidents. Drivers will receive this information through clearly designed in-vehicle displays and warnings.
As part of this European project, the developments will be tested in Austria and Spain over the coming years. The focus will be, among other things, on high-risk situations in tunnels as well as on highway on- and off-ramps. The project will run until March 2030. By then, the technologies must prove themselves in practice. It remains to be seen whether this will actually prevent accidents and how reliable the predictions are.
Warnings must be easy to understand
Two departments at Chemnitz University of Technology are involved in ROAD4SAFE. The team led by psychologist Stefan Brandenburg is examining how the system’s information can be conveyed to drivers in a way that is easy to understand. After all, a warning must be designed so that people can recognize the danger as quickly as possible and react to it.
Legal scholar Dagmar Gesmann-Nuissl and her team are investigating the legal requirements necessary for implementation. After all, this involves a massive amount of data from various sources that is aggregated for risk predictions.
ROAD4SAFE is intended to contribute to the European “Vision Zero". By 2030, the EU aims to halve the number of traffic fatalities and serious injuries compared to 2019. By 2050, the number of traffic fatalities is expected to be close to zero. According to the most recently published figures from the European Commission, a total of 19,350 people died on the roads of the European Union in 2025. Compared to 2019, this represents a 15 percent decrease. The current rate of decline is therefore insufficient to meet the goal of halving the number of fatalities.