Theoretically, to secure and release automated vehicles, an infinite number of test cases must be checked to cover all scenarios of real traffic. The resulting test space is not manageable even with modern methods and simulation tools. One possibility is the scenario-based approach, which only takes the most interesting and relevant scenarios into account to limit the infinite test space to a finite number of test cases. In this thesis, we present an optimization-based approach to generate more complex test scenarios for automated vehicles by means of Evolutionary Algorithm (EA). Tuning experiments with Genetic Algorithm (GA) and Particle Swarm Optimization (PSO) are performed to achieve better optimization performance. Unlike most other methods, our optimization approach is highly related to the actual highway situation and can ensure the physical possibilities of the motion of traffic participants through the defined objective function. It can also keep the relationship between exploration and exploitation in EA well balanced and achieve a sufficient optimization performance with the tuned parameters.     «

Theoretically, to secure and release automated vehicles, an infinite number of test cases must be checked to cover all scenarios of real traffic. The resulting test space is not manageable even with modern methods and simulation tools. One possibility is the scenario-based approach, which only takes the most interesting and relevant scenarios into account to limit the infinite test space to a finite number of test cases. In this thesis, we present an optimization-based approach to generate more...     »

Theoretically, to secure and release automated vehicles, an infinite number of test cases must be checked to cover all scenarios of real traffic. The resulting test space is not manageable even with modern methods and simulation tools. One possibility is the scenario-based approach, which only takes the most interesting and relevant scenarios into account to limit the infinite test space to a finite number of test cases. In this thesis, we present an optimization-based approach to generate more complex test scenarios for automated vehicles by means of Evolutionary Algorithm (EA). Tuning experiments with Genetic Algorithm (GA) and Particle Swarm Optimization (PSO) are performed to achieve better optimization performance. Unlike most other methods, our optimization approach is highly related to the actual highway situation and can ensure the physical possibilities of the motion of traffic participants through the defined objective function. It can also keep the relationship between exploration and exploitation in EA well balanced and achieve a sufficient optimization performance with the tuned parameters.     «

Theoretically, to secure and release automated vehicles, an infinite number of test cases must be checked to cover all scenarios of real traffic. The resulting test space is not manageable even with modern methods and simulation tools. One possibility is the scenario-based approach, which only takes the most interesting and relevant scenarios into account to limit the infinite test space to a finite number of test cases. In this thesis, we present an optimization-based approach to generate more...     »