Empirical Analysis of Driver Behavior at Uncontrolled Midblock Crosswalks: Influence of Bike Lane Buffer Types

Abstract At unsignalized midblock crosswalks (UMCs), visual obstructions created by bike lane buffers can substantially affect drivers’ pedestrian detection and yielding behavior. However, limited evidence is available regarding how different types of bike lane buffers influence driving performance through changes in visual attention and cognitive workload. To address this issue, four representative bike lane buffer scenarios—Lawn, Railing, Bush, and Tree—were developed in a driving simulation environment. Vehicle operation parameters and eye-tracking data were collected simultaneously to investigate drivers’ behavioral and visual responses while approaching a crosswalk. The results showed that highly obstructive buffers significantly delayed pedestrian detection and weakened drivers’ early risk perception. In the Bush and Tree scenarios, pedestrians were detected at distances of only about 18 m from the crosswalk, representing a reduction of more than 40% compared with the low-obstruction conditions. As a consequence, drivers exhibited lower terminal braking speeds, higher maximum deceleration rates, and more abrupt braking maneuvers. The mean braking initiation distance in the Bush and Tree scenarios was approximately 48 m, which was more than 10 m shorter than that observed under the Lawn and Railing conditions. Eye-tracking analyses further revealed that highly obstructive environments significantly increased fixation duration on roadside areas while reducing the effective visual search range. In addition, the proportions of saccades and blinks decreased to approximately 3%, whereas the coefficient of variation of pupil diameter increased to 0.23, indicating elevated cognitive workload and mental stress. These findings demonstrate that visual obstructions associated with bike lane buffers can adversely affect drivers’ hazard perception, visual search behavior, and braking responses. The study provides empirical evidence to support the safety evaluation and design optimization of bike lane buffers near unsignalized midblock crosswalks.