Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Data-Driven Analysis of Crash Patterns in SAE Level 2 and Level 4 Automated Vehicles Using K-means Clustering and Association Rule Mining

Jewel Rana Palit, Vijayalakshmi K Kumarasamy, Osama A. Osman

TL;DR

This study addresses the need to understand crash patterns of SAE Level 2 (ADAS) and Level 4 (ADS) automated vehicles using a nationwide NHTSA dataset. It introduces a two-stage framework that first applies K-means clustering to partition crashes into $K=4$ homogeneous groups based on temporal, spatial, and environmental factors, then uses Association Rule Mining (ARM) to extract interpretable, multivariate crash-contributor relationships within each cluster. Key contributions include leveraging over 2,500 national crashes (post-cleaning: 2380×31), feature engineering of temporal variables, and cluster-specific ARM rules to reveal context-rich patterns that inform developers, regulators, and policymakers. The findings highlight distinct crash regimes for urban Level 4 ADS (intersections) versus highway Level 2 ADAS (front-end impacts), offering actionable guidance for deployment strategies, safety standards, and infrastructure planning.

Abstract

Automated Vehicles (AV) hold potential to reduce or eliminate human driving errors, enhance traffic safety, and support sustainable mobility. Recently, crash data has increasingly revealed that AV behavior can deviate from expected safety outcomes, raising concerns about the technology's safety and operational reliability in mixed traffic environments. While past research has investigated AV crash, most studies rely on small-size California-centered datasets, with a limited focus on understanding crash trends across various SAE Levels of automation. This study analyzes over 2,500 AV crash records from the United States National Highway Traffic Safety Administration (NHTSA), covering SAE Levels 2 and 4, to uncover underlying crash dynamics. A two-stage data mining framework is developed. K-means clustering is first applied to segment crash records into 4 distinct behavioral clusters based on temporal, spatial, and environmental factors. Then, Association Rule Mining (ARM) is used to extract interpretable multivariate relationships between crash patterns and crash contributors including lighting conditions, surface condition, vehicle dynamics, and environmental conditions within each cluster. These insights provide actionable guidance for AV developers, safety regulators, and policymakers in formulating AV deployment strategies and minimizing crash risks.

Data-Driven Analysis of Crash Patterns in SAE Level 2 and Level 4 Automated Vehicles Using K-means Clustering and Association Rule Mining

TL;DR

This study addresses the need to understand crash patterns of SAE Level 2 (ADAS) and Level 4 (ADS) automated vehicles using a nationwide NHTSA dataset. It introduces a two-stage framework that first applies K-means clustering to partition crashes into homogeneous groups based on temporal, spatial, and environmental factors, then uses Association Rule Mining (ARM) to extract interpretable, multivariate crash-contributor relationships within each cluster. Key contributions include leveraging over 2,500 national crashes (post-cleaning: 2380×31), feature engineering of temporal variables, and cluster-specific ARM rules to reveal context-rich patterns that inform developers, regulators, and policymakers. The findings highlight distinct crash regimes for urban Level 4 ADS (intersections) versus highway Level 2 ADAS (front-end impacts), offering actionable guidance for deployment strategies, safety standards, and infrastructure planning.

Abstract

Automated Vehicles (AV) hold potential to reduce or eliminate human driving errors, enhance traffic safety, and support sustainable mobility. Recently, crash data has increasingly revealed that AV behavior can deviate from expected safety outcomes, raising concerns about the technology's safety and operational reliability in mixed traffic environments. While past research has investigated AV crash, most studies rely on small-size California-centered datasets, with a limited focus on understanding crash trends across various SAE Levels of automation. This study analyzes over 2,500 AV crash records from the United States National Highway Traffic Safety Administration (NHTSA), covering SAE Levels 2 and 4, to uncover underlying crash dynamics. A two-stage data mining framework is developed. K-means clustering is first applied to segment crash records into 4 distinct behavioral clusters based on temporal, spatial, and environmental factors. Then, Association Rule Mining (ARM) is used to extract interpretable multivariate relationships between crash patterns and crash contributors including lighting conditions, surface condition, vehicle dynamics, and environmental conditions within each cluster. These insights provide actionable guidance for AV developers, safety regulators, and policymakers in formulating AV deployment strategies and minimizing crash risks.
Paper Structure (14 sections, 4 equations, 7 figures, 6 tables, 1 algorithm)

This paper contains 14 sections, 4 equations, 7 figures, 6 tables, 1 algorithm.

Table of Contents

  1. Abstract
  2. Introduction
  3. Research Gap & Contribution
  4. METHODOLOGY
  5. Data
  6. Data Preparation and Exploratory Analysis
  7. K-means Clustering
  8. Association Rule Mining (ARM)
  9. RESULTS
  10. K-means Clustering
  11. Association Rule Mining(ARM)
  12. DISCUSSION
  13. CONCLUSIONS
  14. AUTHOR CONTRIBUTIONS

Figures (7)

  • Figure 1: AV Crash Analysis Architecture
  • Figure 2: Exploratory Analysis (a) Heatmap of Incident Time, (b)SV Pre Crash Speed Range (mph), (c) Crash % Distribution on Automation Level, (d) Crash % Distribution on Automation System ADS/ADAS, (e) Highest Injury Severity Alleged, (f) AV Crashes for top 10 city
  • Figure 3: Descriptive Statistics of Numeric Variable
  • Figure 4: Descriptive Statistics of Categorical Variable
  • Figure 5: One Hot Encoded Data Size Conversion
  • ...and 2 more figures