Evaluating pedestrian vehicle interaction dynamics at un-signalized intersections: A proactive approach for safety...

Area of Science:

• Traffic Safety Engineering
• Transportation Science
• Data Analytics

Background:

• Unsignalized intersections pose complex safety challenges due to heterogeneous traffic and varied road user behaviors.
• Conventional grid-based analysis methods have limitations in accurately assessing safety risks at these intersections.
• Advanced trajectory data offers potential for more nuanced analysis of road user interactions.

Purpose of the Study:

• To develop and validate an advanced pattern-based approach for categorizing pedestrian-vehicle interactions at unsignalized intersections.
• To improve the estimation of surrogate safety measures (SSM) by incorporating behavioral patterns.
• To propose a framework for evaluating and enhancing the safety of uncontrolled urban intersections.

Main Methods:

• Utilized advanced trajectory-based data for analyzing road user interactions.
• Developed a two-interaction pattern approach categorizing responsive and non-responsive road user behaviors.
• Employed surrogate safety measures (SSM) including Time to Collision (TTC) and Post Encroachment Time (PET) for behavior categorization.
• Applied a k-fold test for variable importance and an Import Vector Machine (IVM) for severity classification.
• Analyzed 1486 interaction events from three unsignalized intersections in India.

Main Results:

• Identified TTC as crucial for pattern-1 (responsive behavior) and both TTC and PET for pattern-2 (non-responsive behavior).
• Successfully classified interaction severity levels using the proposed Import Vector Machine (IVM) model.
• Demonstrated that non-evasive behaviors can also lead to critical interactions, highlighting the need for comprehensive safety analysis.
• The advanced framework provides a more accurate assessment of safety risks compared to traditional methods.

Conclusions:

• The proposed pattern-based approach using trajectory data offers a significant improvement for analyzing road user interactions and safety at unsignalized intersections.
• The developed severity classification system aids in evaluating the effectiveness of safety interventions.
• This research provides a robust framework for improving the safety of uncontrolled urban intersections by understanding complex road user behaviors.