논문 상세보기
pp.255-263
PURPOSES : This study aims to understand the characteristics of accidents involving autonomous vehicles and derive the causes of accidents from road spatial information through autonomous vehicle accident reports. METHODS : For this study, autonomous vehicle accident reports collected and managed by the CA DMV were used as data sources. In addition, spatial characteristics and geometric data for accident locations were extracted by Google maps. Based on the collected data, the study conducted general statistics, text embedding, and cross-analysis to understand the overall characteristics of autonomous vehicle accidents and their relationship with road spatial features. RESULTS : The analysis results for characteristics of autonomous vehicle accidents, applying statistical analysis and text embedding techniques, reveal that the damages caused by autonomous vehicle accidents are often minor, and approximately half of the accidents are triggered by other vehicles. It is noteworthy that accidents where autonomous vehicles are at fault are not uncommon, and when the cause of the accident is within the autonomous vehicle, the accident risk can increase. The accident analysis results using spatial data showed that the severity of accidents increases when on-street parking is present, when dedicated lanes for bicycles and buses exist, and when bus stops are present. CONCLUSIONS : Through this study, geometric and spatial elements that appear to have an impact on autonomous driving systems have been identified. The findings of this study are expected to serve as foundational data for improving the safety of autonomous vehicle operations in the future.
2. 선행 연구
3. 연구 데이터
3.1. 자율주행차 사고 보고서
3.2. 공간정보 데이터
4. 연구 방법론
5. 분석 결과
5.1. 통계 분석
5.2. 텍스트 임베딩
5.3. 기하 데이터 분석
6. 결론
감사의 글
