Modeling Driver Behavior near Intersections in Hidden Markov Model

Area of Science:

• Traffic Safety
• Driver Behavior Analysis
• Computational Modeling

Background:

• Intersections are critical safety zones with frequent driver behavior changes leading to collisions.
• Understanding driver behavior, particularly in the 'dilemma zone,' is crucial for accident prevention.

Purpose of the Study:

• To analyze driver behavior at intersections using a Hidden Markov Model (HMM).
• To quantify driver stability and risk, especially within the dilemma zone.
• To develop a model for predicting potential dangers and enhancing driving assistance systems.

Main Methods:

• Utilized discrete data processing of observed vehicle dynamics.
• Applied the Baum-Welch (B-W) estimation algorithm for HMM parameter calculation.
• Employed the Forward algorithm to determine the most likely driver states.
• Incorporated the Viterbi Algorithm for potential danger prediction.

Main Results:

• Driver behavior in the dilemma zone demonstrates significantly lower stability and higher risk.
• The HMM successfully models and measures the stability and risk of driver behavior.
• The Viterbi Algorithm effectively predicts potential vehicle dangers.

Conclusions:

• Driver behavior in the dilemma zone poses a higher safety risk.
• The developed HMM provides a robust framework for analyzing intersection driver behavior.
• Findings can inform the development of advanced driving assistance systems for accident mitigation.