Scholarlink

In order to improve road safety, recent studies suggest that it is important to study and identify the optimal driving benchmarks that reflect the safest driving behaviour that may be observed by human drivers. The objective of this paper is to identify boundaries of risky and typical driving by studying the car-following driving behaviour. The data used in this study was collected by TNO in a recent naturalistic driving study. The distributions of driving metrics related to the following and leading vehicle were illustrated to understand their shapes and outliers. The safety-related car-following driving metrics of Time to Collision (TTC), Deceleration Rate to Avoid the Crash (DRAC), Crash Index (CI) and over-speeding were calculated, with risky thresholds obtained from the literature, and typical driving thresholds based expert assessors’ ratings. Principal Component Analysis (PCA) was applied to these metrics and showed that ‘optimal driving’ can be represented by one linear component that represents over 95% of the total dataset’s variance.

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

Can Safe Driving Patterns Be Identified? An Exploratory Analysis

  • 会议论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年

查看全文
  • Dimitrios I. Tselentis,
  • Eleonora Papadimitriou,
  • Arturo Tejada

摘要

In order to improve road safety, recent studies suggest that it is important to study and identify the optimal driving benchmarks that reflect the safest driving behaviour that may be observed by human drivers. The objective of this paper is to identify boundaries of risky and typical driving by studying the car-following driving behaviour. The data used in this study was collected by TNO in a recent naturalistic driving study. The distributions of driving metrics related to the following and leading vehicle were illustrated to understand their shapes and outliers. The safety-related car-following driving metrics of Time to Collision (TTC), Deceleration Rate to Avoid the Crash (DRAC), Crash Index (CI) and over-speeding were calculated, with risky thresholds obtained from the literature, and typical driving thresholds based expert assessors’ ratings. Principal Component Analysis (PCA) was applied to these metrics and showed that ‘optimal driving’ can be represented by one linear component that represents over 95% of the total dataset’s variance.