Event-Based Safety Indicator Analysis of a Surround View Monitoring (SVM)-Based Pedestrian Detection System Using...

Area of Science:

• Road Safety Engineering
• Human-Computer Interaction
• Automotive Technology

Background:

• Urban pedestrian safety is a significant challenge, exacerbated by vehicle blind spots.
• Surround View Monitoring (SVM) systems offer potential for enhanced situational awareness.

Purpose of the Study:

• To investigate the relationship between SVM-defined risk zones and safety indicators.
• To assess the utility of SVM in identifying and quantifying driving risks in pedestrian-heavy areas.

Main Methods:

• Utilized real-world driving data from school zones and crash-prone areas in Seoul.
• Collected vehicle diagnostics, RTK positioning, SVM outputs, and video data.
• Reconstructed vehicle-vulnerable road user interactions and analyzed safety metrics like Time to Collision (TTC) and jerk.

Main Results:

• SVM-defined danger zones exhibited lower TTC and higher Time-Integrated TTC (TIT)/Time-Exposed TTC (TET) ratios, indicating greater immediate risk.
• Warning zones showed higher longitudinal jerk, suggesting more frequent vehicle control adjustments.
• A decrease in 95th percentile total jerk in danger zones over repeated drives was observed but deemed exploratory.

Conclusions:

• SVM-defined risk zones correlate with distinct event-level risks and vehicle dynamic behaviors.
• SVM shows promise as a tool for event-based safety assessments in urban driving environments.