Collision risk identification and prediction considering heterogeneous braking patterns using large-scale...

Area of Science:

• Road safety engineering
• Vehicle dynamics
• Traffic accident analysis

Background:

• Rear-end collisions are a common traffic safety issue, often resulting from inadequate vehicle deceleration during braking.
• Existing safety measures may not fully capture the complex dynamics leading to these crashes.

Purpose of the Study:

• To develop and validate a novel method for identifying and predicting collision risk in pre-crash scenarios.
• To integrate the braking dynamics of both leading and following vehicles for enhanced risk assessment.

Main Methods:

• Utilized a piecewise linear model for deceleration profiles and identified 45 collision risk moments across 10 scenarios.
• Proposed a critical time-to-collision metric incorporating braking timing and execution.
• Employed Gaussian mixture regression to model driver heterogeneity and generate interval-valued crash risk predictions.

Main Results:

• Collision risk moments were found to be dependent on the braking timing and execution of both vehicles.
• The novel critical time-to-collision metric demonstrated superior performance over traditional surrogate safety measures.
• The proposed metric achieved higher accuracy and lower variability in predicting collision risk.

Conclusions:

• The critical time-to-collision metric provides a reliable and effective measure for collision risk prediction.
• Findings support the integration of this metric into collision avoidance systems and advanced driver assistance technologies.
• This research contributes to improving road safety by offering a more accurate approach to anticipating and preventing rear-end collisions.