Estimation of red-light running frequency using high-resolution traffic and signal data

Area of Science:

• Traffic Engineering
• Road Safety
• Transportation Science

Background:

• Red-light-running (RLR) significantly contributes to intersection-related crashes and compromises road safety.
• Understanding the factors influencing RLR and its frequency is crucial for developing effective countermeasures.
• Existing methods often rely on video analysis, necessitating alternative approaches for large-scale data collection.

Purpose of the Study:

• To identify influential factors associated with red-light-running (RLR).
• To estimate RLR frequency using traffic and signal event data without video surveillance.
• To develop a predictive model for RLR occurrence.

Main Methods:

• Utilized high-resolution traffic and signal event data from loop detectors at five intersections.
• Developed a novel method to identify RLR events using stop bar, downstream, and advance detector data.
• Analyzed 6550 identified RLR cases over 12 months to investigate relationships with traffic, signal, and environmental factors.

Main Results:

• RLR frequency is influenced by factors such as arrival time, speed, headway, traffic demand, signal cycle length, intersection geometry, and weather.
• RLR is most prevalent during weekdays, peak periods, under high traffic volumes, and with longer signal cycles.
• 95.24% of RLR events occurred within the first 1.5 seconds of the red phase, particularly during phase transitions and with short headways.

Conclusions:

• Vehicles are more likely to run red lights when approaching intersections during phase transitions and when following closely behind other vehicles.
• A simplified nonlinear regression model can effectively estimate RLR frequency based on advance detector data.
• Findings provide valuable insights for improving intersection safety and reducing RLR violations.