Scholarlink

<p>This study proposes an integrated real-vehicle–simulation-based methodology for quantitatively evaluating the collision avoidance performance of vehicles equipped with advanced driver assistance systems (ADAS) in highway cut-in scenarios. Repeatable real-vehicle tests are conducted on two production vehicles, the Kia K8 and the Tesla Model 3, using a Target Vehicle that combines the Kookmin University Target Robot (KTR) and a Global Vehicle Target (GVT), and vehicle dynamic data are collected during the period immediately preceding system activation. Experimental results show that both vehicles exhibit collision avoidance behavior through direct intervention of the automatic emergency braking (AEB) system; however, differences are observed in the criteria for assessing the lane intrusion level of the cut-in vehicle and in the timing of AEB activation. Based on the collected data, the lane intrusion detection range and the vehicle intrusion point are defined as key parameters for describing collision risk, and these parameters are incorporated into a high-fidelity simulation model developed in a Carmaker–MATLAB/Simulink co-simulation environment. The developed model is calibrated and validated using real-vehicle test data, achieving RMSE values of 1.69&#xa0;km/h and 1.66&#xa0;km/h for the Tesla Model 3 and the Kia K8 platforms, respectively, demonstrating a high level of fidelity. The proposed integrated framework effectively combines experiment-based ADAS evaluation with scenario-based simulation, enabling accurate reproduction of real-world collision avoidance behavior of production ADAS in high-risk highway cut-in situations, and provides a foundational tool for future autonomous driving and safety function development, scenario-based validation, and regulatory testing.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Scenario-Based Evaluation Framework for AEB Systems Using Real Vehicle Tests and Simulation in Highway Cut-in Situation

  • 期刊论文

  • 发表日期：

    2026年5月12日

查看全文
  • Myungyeun Park,
  • Minjong Choi,
  • Jayil Jeong

摘要

This study proposes an integrated real-vehicle–simulation-based methodology for quantitatively evaluating the collision avoidance performance of vehicles equipped with advanced driver assistance systems (ADAS) in highway cut-in scenarios. Repeatable real-vehicle tests are conducted on two production vehicles, the Kia K8 and the Tesla Model 3, using a Target Vehicle that combines the Kookmin University Target Robot (KTR) and a Global Vehicle Target (GVT), and vehicle dynamic data are collected during the period immediately preceding system activation. Experimental results show that both vehicles exhibit collision avoidance behavior through direct intervention of the automatic emergency braking (AEB) system; however, differences are observed in the criteria for assessing the lane intrusion level of the cut-in vehicle and in the timing of AEB activation. Based on the collected data, the lane intrusion detection range and the vehicle intrusion point are defined as key parameters for describing collision risk, and these parameters are incorporated into a high-fidelity simulation model developed in a Carmaker–MATLAB/Simulink co-simulation environment. The developed model is calibrated and validated using real-vehicle test data, achieving RMSE values of 1.69 km/h and 1.66 km/h for the Tesla Model 3 and the Kia K8 platforms, respectively, demonstrating a high level of fidelity. The proposed integrated framework effectively combines experiment-based ADAS evaluation with scenario-based simulation, enabling accurate reproduction of real-world collision avoidance behavior of production ADAS in high-risk highway cut-in situations, and provides a foundational tool for future autonomous driving and safety function development, scenario-based validation, and regulatory testing.