Scholarlink

<p>To address the limitations of current in-depth investigation systems for intelligent connected vehicle accidents, this paper analyzes the complexity and characteristics of modern self-driving crashes using the 2021 Annual Report on Intelligent Connected Vehicles published by the California Department of Motor Vehicles. A closed-loop framework is proposed, integrating human–vehicle–road collaborative data collection, reverse scenario reconstruction, and accident causation analysis. At the technical level, high-precision, time-sequenced motion parameters are extracted from vehicle log data. A two-wheel vehicle dynamics model is established, and a 20-second pre-crash spatiotemporal trajectory reconstruction is developed using MATLAB. OpenStreetMap (OSM) data is integrated and converted to the Open DRIVE format via SUMO tools, enabling accurate reconstruction of road geometry. To handle heterogeneous multimodal log data, Python-based algorithms are employed for data cleaning and feature extraction, capturing lateral vehicle displacement and surrounding traffic dynamics. Finally, a parameterized model of the pre-collision dynamic scene is built using the Open SCENARIO standard, achieving high-fidelity accident reproduction. This method offers scientific support for liability determination, safety enhancement of autonomous systems, and the development of standardized evaluation frameworks.</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 study on in-depth accident investigation of intelligent connected vehicles based on reverse scenario construction

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年11月21日

查看全文
  • Shibo Zhang,
  • Xin Li,
  • Zhiyong Zha,
  • Huiwen Xiong,
  • Chengcheng Wu,
  • Pingfei Li,
  • Zhenyu Kang

摘要

To address the limitations of current in-depth investigation systems for intelligent connected vehicle accidents, this paper analyzes the complexity and characteristics of modern self-driving crashes using the 2021 Annual Report on Intelligent Connected Vehicles published by the California Department of Motor Vehicles. A closed-loop framework is proposed, integrating human–vehicle–road collaborative data collection, reverse scenario reconstruction, and accident causation analysis. At the technical level, high-precision, time-sequenced motion parameters are extracted from vehicle log data. A two-wheel vehicle dynamics model is established, and a 20-second pre-crash spatiotemporal trajectory reconstruction is developed using MATLAB. OpenStreetMap (OSM) data is integrated and converted to the Open DRIVE format via SUMO tools, enabling accurate reconstruction of road geometry. To handle heterogeneous multimodal log data, Python-based algorithms are employed for data cleaning and feature extraction, capturing lateral vehicle displacement and surrounding traffic dynamics. Finally, a parameterized model of the pre-collision dynamic scene is built using the Open SCENARIO standard, achieving high-fidelity accident reproduction. This method offers scientific support for liability determination, safety enhancement of autonomous systems, and the development of standardized evaluation frameworks.