Scenario Parameter Generation Method and Scenario Representativeness Metric for Scenario-Based Assessment of Automated Vehicles

The development of assessment methods for the performance Automated Vehicles (AVs) is essential to enable deployment automated driving technologies, due complex operational domain AVs. One candidate scenario-based assessment, in which test cases are derived from real-world road traffic scenarios obtained data. Because high variety possible scenarios, using only observed not sufficient. Therefore, generating additional necessary. Our contribution twofold. First, we propose a method determine parameters that describe sufficient degree while relying less on strong assumptions characterize scenarios. By estimating probability density function (pdf) these parameters, realistic parameter values can be generated. Second, present Scenario Representativeness (SR) metric based Wasserstein distance, quantifies what extent with generated representative covering actual found A comparison our proposed scenario parameterization and pdf estimation shows automatically optimal estimation. Furthermore, it demonstrated SR used choose (number of) best scenario. presented promising, because directly applied already available importance sampling strategies accelerating evaluation