| With the continuous improvement of vehicle intelligence,it is increasingly difficult for traditional open road testing to meet the reliability and robustness requirements of autonomous vehicles.The scenario-based simulation test method has huge technical advantages in terms of test efficiency and test cost,and has gradually become an important means of testing and verification of autonomous vehicles.Efficient and comprehensive scenario testing requires rigorous,flexible and standardized scenario representation to support.However,the current scenario description languages still have the following problems:(1)The language is often related to specific tools and has a low degree of abstraction;(2)The existing scenario description languages are difficult to accurately describe the complex behavior sequences of scenario participants,and the execution process has great uncertainty.Aiming at the above problems,this thesis designs and implements the dynamic scenario description language X-Scenic.The language is based on the scenario description language Scenic with a high degree of abstraction and strong static scenario description ability,and expands the dynamic scenario description ability of the language.In order to support the characterization of complex dynamic test scenarios,the thesis designs a phase-based dynamic behavior description method for scenario participants,which can reduce the uncertainty during scenario execution and improve the success rate of scenario execution as expected.At the same time,X-Scenic adopts a specific scenario generation method based on constraint solving,which improves the efficiency of test scenario generation.The experimental results show that X-Scenic not only enhances the ability of dynamic scenario description,but also significantly reduces the time for generating specific scenarios and improves the efficiency of automated driving simulation tests.Based on the research on the above-mentioned key technologies,this thesis further designs and implements an autonomous driving simulation test tool based on X-Scenic,which provides functions such as scenario management and parameter configuration,which can help testers perform simulation tests efficiently,thereby improving the research and development efficiency of autonomous driving systems. |