Sensing walkability: Insights from wearable sensors on pedestrian stress and safety

Walking is recognized for its benefits to sustainability, public health, and equity. A critical factor affecting the adoption of walking is travel stress, as pedestrians are less likely to choose walking or other active modes of transportation when they feel unsafe or stressed in their surrounding environment. Although the importance of pedestrian stress has been increasingly recognized, there are many important questions that remain to be answered. For example, how to measure the nuanced, moment-to-moment fluctuations in stress that occur as pedestrians move through different urban environments? How to evaluate the impact of the micro-level built environment on pedestrians while walking? The advancement of wearable, mobile sensors offer promising opportunities to address these questions and collect real-time, high-resolution data on pedestrian stress and behavior. Devices that track physiological responses, such as heart rate variability (HRV), electrodermal activity (EDA), or eye movement, can provide direct insight into pedestrians’ emotional and cognitive states as they navigate the built environment. By leveraging such tools, we can better understand how pedestrians perceive and respond to stressors, ultimately informing the design of safer, more inclusive urban spaces. In this dissertation, I present three studies that evaluate the use of sensors to examine pedestrians’ safety and comfort. In the first study, I test the feasibility of the methodology, investigating whether and how biometric sensors can be used to identify pedestrian stress levels. I compare the extent to which self-reported stress collected via a survey aligns with 48 sensor-measured stress metrics, highlighting the time-sensitivity of different sensor measurements and the importance of handling individual differences of sensor signals in the analytical method. In the second study, I reflect on the limitations of current gaze metrics, and propose an innovative metric for visual attention that accounts for both fixation duration and object size. By comparing the results of visual attention during intersections, this new metric stands out as a better indicator for AOI importance and pedestrian alertness. In the third and final study, I examine how perceptions of the street-level built environment affect pedestrian stress using eye-tracking data. Applying the knowledge from the previous two studies, both built environment features and stress indicators are extracted from the eye-tracking data. The findings provide empirical evidence for urban design concepts and offer actionable insights for pedestrian-friendly street planning.