Although physical visibility is often assumed to ensure driver awareness, relatively little naturalistic evidence describes how drivers perceive vulnerable road users (VRUs) across real-world driving contexts. This study examined roadway and driver factors associated with detection of pedestrians, cyclists, and motorcyclists using synchronized multimodal data. Twenty-four participants completed two naturalistic drives over a fixed route representing rural, suburban, campus, and urban environments. Driver gaze was recorded using Tobii Pro Glasses 3, vehicle data using a CANbus/GNSS-IMU logger, and heart rate using a Polar H10 sensor. The synchronized dataset contained 3,526 annotated VRU encounters. Mixed effects logistic regression identified factors associated with successful detection. Vehicle speed was the strongest factor, with each 1-mph increase associated with an approximately 5% reduction in detection odds (OR = 0.95, p < 0.001). VRUs on the right side of the roadway were twice as likely to be detected as those on the left (OR = 2.00, p < 0.001), and detection was higher in rural than urban environments (OR = 4.55, p = 0.019). Findings support speed management, context-sensitive roadway design, and driver-assistance systems to improve VRU safety.

Infrastructure Assets: Highway Assets
Resource Types: Research Report
Capabilities: Data & Information Systems
Management Processes: Strategic Direction