For this purpose, they installed around 80,000 square meters of variable-use asphalt surfaces. The system includes lanes, large areas, intersection areas, and even tram tracks. With flexible markings, mobile infrastructure, and situational perimeter development, it is now possible to display a wide variety of scenarios for inner-city and, in some cases, intercity traffic. The concrete layout of the asphalt surfaces of the new city courses was planned by DEKRA in a joint working group with vehicle manufacturers, suppliers, and research institutions. “Under the leadership of the Fraunhofer Institute for Traffic and Infrastructure Systems in Dresden, we evaluated the inner-city and suburban accident history in Germany from 2013 to 2019 in detail,” explains Uwe Burckhardt, Head of Test and Event at DEKRA Lausitzring. With the road geometries of the flexible city courses, more than 80 percent of these real traffic scenarios can be reproduced in the test.
Reproducibility in tests can validate simulations
“Using state-of-the-art technology, we can put vehicles in almost any complex situation to expose their automated driving functions to the maximum test challenge,” Burckhardt goes into detail. In so-called swarm tests, up to twelve moving objects could be used in the test vehicle’s environment. For this purpose, data exchange between the control station and the targets must take place in real time or with very low latency – preferably via the 5G communication standard. These objects can represent a wide variety of other vehicles as well as pedestrians and can be controlled with centimeter precision by using, among other things, the Differential Global Positioning System. This means that the same test sequences can be reproduced again and again, so that systems and functions can be tested under the same conditions in each case. For example, the reaction of an emergency brake assistant to the end of a traffic jam or a pedestrian suddenly crossing the road, the behavior of a turn-off assistant with a cyclist in the driver’s blind spot, or the reliability of highly and fully automated and autonomous systems, to name just a few examples.
Reproducibility is, in fact, a very important keyword in this context. “In order to achieve representative and statistically reliable results in endurance tests in real traffic, it’s necessary to drive distances of many millions of kilometers,” explains Burckhardt. And cameras and sensors have to prove themselves in around 15,000 individual tests before they can be installed in a production vehicle. This is hardly economically feasible. Therefore, as many situations as possible have to be simulated in computer systems under a wide range of boundary conditions. The better the simulation systems are matched to real conditions, the better the results they deliver. It must also be possible to recreate situations that produce critical results in the simulation as reproducibly as possible in reality.
That is why it is a great advantage for developers in the automotive and supplier industry if such simulations can be carried out on the basis of specific road data, on which the tests are later also reproduced in practice. And exactly this is possible with the DEKRA Lausitzring’s terrain model and surrounding test tracks. For this purpose, DEKRA had the entire test site extensively digitized using a 3D laser scan. Based on this, a 3D terrain model was developed, which optimally serves the specific planning of tests to validate functions of highly automated driving.