Related Experiment Video
Updated: Jul 15, 2026
08:22
A Novel Single Animal Motor Function Tracking System Using Simple, Readily Available Software
Published on: August 31, 2018
6.5K
Cyclist safety assessment using autonomous vehicles
1Department of Civil Engineering, The University of British Columbia, Canada.
Accident; Analysis and Prevention
|January 21, 2025
Summary
Prioritizing cyclist safety, this study uses autonomous vehicle data to estimate real-time crash risk. A new method identifies high-risk areas, enabling safer route planning for cyclists.
Area of Science:
- Transportation Engineering
- Road Safety Management
- Traffic Analysis
Background:
- Vehicle-cyclist crashes are high-severity, necessitating proactive safety management.
- Quantifying cyclist crash risk is challenging due to sparse collision data.
- Advanced technologies are needed for proactive, multi-modal road safety.
Purpose of the Study:
- To develop a conflict-based methodology for estimating dynamic cyclist crash risk.
- To leverage autonomous vehicle data for real-time safety assessment.
- To enable proactive, route-level safety metrics for cyclists.
Main Methods:
- Utilized 87 hours of autonomous vehicle data (nuPlan) from downtown Boston.
- Identified traffic conflicts to extrapolate crash risk.
- Developed a Bayesian Hierarchical Extreme Value model for dynamic risk assessment.
Main Results:
- Estimated real-time crash risk for intersections and mid-blocks.
- Found cyclist facilities generally safer than shared facilities, but with significant temporal variations.
- Observed instances where shared facilities were safer than designated lanes, emphasizing real-time needs.
Conclusions:
- A conflict-based approach using AV data can estimate dynamic cyclist crash risk.
- Real-time safety monitoring is crucial due to fluctuating risk levels.
- A user-level application for safest route planning based on real-time risk is feasible.
