TY  - JOUR
PY  - 2022//
TI  - Accident risk prediction of intelligent vehicle takeover process at urban roundabout
JO  - China safety science journal (CSSJ)
A1  - Liu, Q.
A1  - Xu, T.
A1  - Xiong, X.
A1  - Zhao, J.
A1  - Cai, Y.
SP  - 150
EP  - 157
VL  - 32
IS  - 12
N2  - Aiming at the accident risk in the process of intelligent vehicle takeover at the urban roundabout, a prediction model of intelligent vehicle accident risk at the roundabout was proposed to reduce the accident rate. Based on the urban traffic simulation software simulation of urban mobility (SUMO), the scene of urban roundabout was established, and the accident data of intelligent driving vehicle takeover process was analyzed to reveal the impact mechanism of road area and takeover time (ToT) on the accident risk of intelligent vehicle takeover process. CatBoost was used to model the accident risk, and the sensitivity analysis and Area Under the Curve (AUC) were used as evaluation indicators to compare the prediction performance with Linear Regression and XGBoost models. The results show that the impact of speed on the accident is more than 47%. The accident rate in the entrance lane of the intersection is the highest. The accident rate of the left lane of the entrance is 8. 63% higher than that of the right lane on average. The influence of (ToT) on car accident rate at roundabouts is about 8. 5%, and the proportion of road curvature and radius factors in the loop area on accident factors is less than 5%. The roundabout segment hardly affects the autonomous vehicle collision during the vehicle takeover process. The prediction accuracy of the CatBoost model is higher than that of linear regression and XGBoost. © 2022 China Safety Science Journal. All rights reserved.<p /> <p>Language: zh</p>
LA  - zh
SN  - 1003-3033
UR  - http://dx.doi.org/10.16265/j.cnki.issn1003-3033.2022.12.2650
ID  - ref1
ER  -