This study analyzes the accuracy of vehicle behavior reconstruction based on Data Storage System for Automated Driving (DSSAD) data to ensure the reliability of autonomous vehicle accident analysis. To this end, an experimental vehicle capable of integrating Controller Area Network (CAN) data, Global Navigation Satellite System (GNSS) information, and video data was constructed to establish a simulated environment for the recording system. For precision validation of the reconstruction results, the actual driving trajectory derived from LiDAR and video analysis was set as the reference data. Based on this, three scenarios were formulated: GNSS-based reconstruction, vehicle speed and steering wheel angle-based reconstruction, and speed and steering wheel angle-based reconstruction incorporating the measured steering gear ratio, with trajectory errors quantitatively evaluated for each. Comparative analysis of the nearest-distance error and lateral trajectory error revealed that the GNSS-based reconstruction yielded the lowest error, while the combination of vehicle speed and steering wheel angle using default parameters produced the largest error. The findings of this study suggest that, in future accident analyses utilizing DSSAD data, validating the integrity of satellite data and correcting for vehicle-specific dynamic characteristics are key factors in improving reconstruction accuracy.
목차
Abstract 1. 서론 1.1 연구 개요 1.2 선행연구 고찰 2. 이론적 배경 2.1 DSSAD 2.2 EDR 2.3 회전반경 2.4 조향기어비 2.5 PC-Crash 2.6 차량 궤적 정량 평가를 위한 오차 지표 3. 조향 보정 모델 개발 3.1 가변 조향기어비 계산 함수 3.2 Ackermann 기하학 보정 3.3 EDR 데이터 처리 및 PC-Crash 연동 4. 실차시험을 통한 검증 4.1 얼라인먼트 조정 4.2 조향기어비 실측 4.3 주행시험 데이터 수집 4.4 시뮬레이션 구성 및 검증 시나리오 4.5 검증 결과 분석 5. 결론 및 향후 과제 후기 References