面向自动驾驶的系统级安全违规场景生成研究进展OA

The safety of autonomous driving systems is a fundamental requirement for vehicle deployment,making com-prehensive and systematic testing essential.Simulation testing reduces the need for costly real-world evaluations.While autonomous driving systems typically comprise perception,decision-making,and control modules,the existing reviews are largely confined to module-level testing,lacking a systematic treatment of system-level methodologies.To enable researchers to gain a comprehensive understanding of safety violation scenarios across the entire autonomous driving sys-tem,this paper offers a systematic review of safety violation scenario generation at the system level.A scenario derivation framework is proposed,beginning with initial scenario acquisition and proceeding to safety violation generation.Key tech-nical approaches are summarized,and available scenario datasets and simulation tools are compiled.Major challenges in system-level testing are discussed,and future research directions are outlined.