Artificial intelligence assists with accident data analysis
Since June 2019, the German Aerospace Center (Deutsches Zentrum für Luft- und Raumfahrt; DLR) has been working with SSP Consult – Beratende Ingenieure GmbH to develop a system to support road safety work using artificial intelligence (AI). The project will run for two years. The system will use novel machine learning algorithms to evaluate data in order to predict accident frequencies and their influencing factors at junctions. The KI4Safety project will receive a total of 279,000 euro in funding from the German Federal Ministry of Transport and Digital Infrastructure (Bundesministerium für Verkehr und digitale Infrastruktur; BMVI) as part of their mFUND (mobility fund) funding directive.
The data evaluated by the new system using deep learning processes include accident data and expert interviews with the police, accident commissions and planners from a range of different institutions. Aerial images, maps and traffic data are also used. “We use the collected data to define various scenarios for training the algorithm,” explains Andreas Leich, AI4Safety Project Manager at the DLR Institute of Transportation Systems. “At the end we will have a system that supports urban planners or accident commissions in predicting accident black spots within specific areas of road infrastructure and to identify suitable mitigation measures.” Artificial intelligence thus contributes towards a better understanding of the numerous factors that affect the occurrence of traffic accidents and thereby increases road safety.
About the BMVI mFUND
As part of its mFUND research initiative, the BMVI has been supporting research and development projects relating to data-based digital applications for Mobility 4.0 since 2016. In addition to this financial support, mFUND also offers a range of event formats to promote networking between interested parties from government, industry and research, as well as access to the mCLOUD data portal. More information is available on the BMVI website.