When feeling safe becomes risky: A VR-EEG-computer vision framework for analyzing cyclist safety in dynamic traffic...

Area of Science:

• Road safety research
• Human factors in transportation
• Neuroscience of perception

Background:

• The discrepancy between cyclists' perceived safety and actual crash risk in mixed traffic is poorly understood.
• Existing research often neglects the impact of dynamic visual elements on risk perception.

Purpose of the Study:

• To investigate the mismatch between perceived safety and crash risk in cyclists.
• To explore the influence of built environment (BE) features on this perceptual mismatch.

Main Methods:

• Developed a virtual reality (VR) platform integrating electroencephalography (EEG), BE data, and time-to-collision (TTC) from VISSIM.
• Utilized a Long Short-Term Memory (LSTM) model to analyze EEG and environmental data from 72 participants.

Main Results:

• Identified a 'perceptual relief period' where cyclists underestimate risk post-overtaking.
• Found that high spatial enclosure, complex textures, dense vegetation, and low visible vehicle density amplify the risk perception mismatch.
• Demonstrated that aesthetically focused BE features may unintentionally increase risk in high-conflict zones.

Conclusions:

• The study provides empirical evidence for an integrated human-vehicle-environment safety framework.
• Highlights the need for interdisciplinary collaboration between neuroscience and transport engineering for safer cycling infrastructure.
• Suggests that current road design may inadvertently compromise cyclist safety by affecting risk perception.