Despite a fall in the number of accidents, over 3,000 people still die every year in road traffic in Germany. Further reducing this figure and diminishing the severity of accidents are in the interests of society at large; this is all now coming a step closer to reality thanks to autonomous driving. Scientists from the Fraunhofer Institute for High-Speed Dynamics, Ernst-Mach-Institut (EMI), are working on traffic simulations capable of forecasting, which take into account the increasing complexity of road traffic by means of automated driving functions.
Future traffic scenarios will involve a mix of conventional and autonomous vehicles. Testing the latter with a view to reducing accident figures to a minimum entails considerable effort in terms of time and costs: An autonomous vehicle should statistically cover more than 210 billion kilometers, for example, to demonstrate that it causes only half as many accidents as a vehicle driven by a human. A research project promoted by the Daimler and Benz Foundation will provide assistance here. “Traffic simulation model as a basis for forecasting road accident statistics in future traffic scenarios” is the title of a research project at Fraunhofer EMI in Freiburg. This doctoral project involves three points of emphasis: The scientists intend to compile a database that provides precise statistics on real traffic and constitutes a valid basis for simulation models of all traffic scenarios. In parallel, existing simulation models are examined and further developed as required. By the end of the project period, a semi-automated optimization process is to be developed that statistically compares real and simulated traffic data, and on this basis optimizes the traffic flow simulation model to enable statistically valid, holistic simulation of real traffic. It would then be possible for the first time to predict the probability of accidents both in conventional traffic and in mixed traffic with autonomous vehicles reliably and efficiently in terms of time and costs, alone on the basis of observation and analysis of traffic scenarios.