A two-dimensional surrogate safety measure based on fuzzy logic model

Yueru Xu,Wei Ye,Yuanchang Xie,Chen Wang

DOI: https://doi.org/10.1016/j.aap.2024.107529

Abstract:Surrogate Safety Measures (SSM) are extensively applied in safety analysis and design of active vehicle safety systems. However, most existing SSM focus only on the one-dimensional interactions along the vehicle traveling direction and cannot handle the crash risks associated with vehicle lateral movements such as sideswipes and angle crashes. To bridge this important knowledge gap, this study proposes a two-dimensional SSM defined based on Fuzzy Logic and the Inverse Time to Collision (FL-iTTC), which accounts for neighboring vehicles' lateral kinematics and the uncertainty of their movements. The proposed FL-iTTC are proven to be more accurate than traditional SSM in identifying typical risky scenarios, including harsh decelerations, sudden lane-changes, cut-ins and pre-crashes that are extracted from the NGSIM dataset. Additionally, other naturalistic driving scenarios are extracted from the NGSIM dataset and are used to evaluate the effectiveness of different SSM in quantifying crash risks. FL-iTTC is compared with other two-dimensional SSM including Anticipated Collision Time (ACT) and Probabilistic Driving Risk Field (PDRF) based on the confusion matrix and the receiver operating characteristic (ROC) curve. The Area under the ROC Curve (AUC) is 0.923 for FL-iTTC, while only 0.891 for ACT and 0.907 for PDRF, which indicates FL-iTTC outperforms other two-dimensional SSM in risk assessment. Overall, the proposed FL-iTTC greatly complements existing SSM and provides a reliable and useful tool to evaluate various crash risks associated with vehicle lateral movements such as cut-in and sideswipe.