In the development of innovative systems in the sector of integral vehicle safety, a large number of driving tests are needed according to the current state of the art for application and validation. The associated effort is already considerable today and will even increase dramatically in the future in view of the increasingly complex and increasingly integrated networked systems, which among other things involve autonomous driving.
This circumstance is problematic because drives on test tracks, proving grounds and in real traffic events are usually associated with a high expenditure of time and costs. For this reason, an increase of share of computer-aided (virtual) driving tests in the form of HIL and SIL simulations is ultimately inevitable in the future. This topic is of great relevance, especially for vehicle safety systems, because it is the only way to realize a safe market introduction of highly automated vehicles. For an effective use in vehicle safety functions or even in whole vehicles, the quality and informative value of the simulations may not differ significantly from those in the real test on the test track or on public roads. For this reason, a consistent concept should be developed. The steps are shown for an emergency evasion function.
First of all, prediction models for road users, in particular for describing the dynamics of the vehicles, are to be developed. For meaningful models, an accurate knowledge of the real parameters is required. These can only be identified from measurements on the real vehicle. The aim is to obtain all necessary information from a few characteristic experiments. With the fully parameterized models, a variety of driving maneuvers are to be performed in order to assess the driving behavior in as many critical situations as possible.
In addition to the validation of the prediction models, the safety function itself has to be tested. First, the functionality is to be demonstrated in standard maneuvers in simulation and in real vehicles. These test scenarios can be obtained from accident databases such as GIDAS, from hazardous situations in field tests from EuroFOT or from consumer protection guidelines such as EuroNCAP.
The validated simulation model can then be used to perform virtual tests in as many critical situations as possible. For this reason, a traffic simulation is built up in SUMO that includes a variety of complex scenarios involving multiple road users. The various road users are supposed to move randomly on the streets, which are created based on digital maps. Situations that trigger the vehicle safety system or are classified as critical are analyzed in more detail in CARMAKER. Based on the results, the safety function can be evaluated with objective criteria.