Model Based Generation of Driving Scenarios

Abstract

For the system test of automotive safety systems, thousands of kilometers need to be driven on real roads. In the future, that number will increase significantly through higher complexity of the functions. To reduce that number and guarantee the controllability, reproducibility and increase the flexibility, a high amount of virtual driving kilometers will be done in X-in-the-Loop (XiL) tests, simulating sensors, weather conditions, vehicle dynamics, car drivers, vulnerable road users, etc. Defining these driving scenarios manually is very complex, time consuming and can not be traced to test coverage conditions. This paper presents an approach to extract simulation based driving scenarios from state based test models. Through building a test model of the requirements and expending that with scenery and maneuver information of the driving tests, it is shown, that complete driving scenarios can be generated automatically to reach every possible state of the system under test.