Bicyclist overtaking in naturalistic riding data: Speed, clearance, and implications for injury risk.
Abstract
Objective
Motor vehicle passing speed and lateral clearance jointly determine both bicyclist injury risk and perceived safety during overtaking events. Most prior studies have examined these factors separately and/or in limited settings. Real-world evidence describing how passing speed and clearance interact across heterogeneous roadway contexts remains limited. This study evaluates high-resolution naturalistic overtaking data within a kinetic-energy injury framework to evaluate whether roadway context and bicycle infrastructure modify both passing speed and lateral clearance, and whether spatial compensation meaningfully offsets energetic exposure.
Methods
Following institutional review board approval, naturalistic cycling data were collected from six commuter cyclists riding as normal using a sensor suite including cameras, lidar, rear-facing radar, GPS, and inertial measurement unit. Approximately 9,900 km of riding were recorded over four months. Overtaking events were detected and characterized using time-synchronized radar and lidar data, yielding 8,753 passes with reliable vehicle speed estimates after quality filtering. Overtaking events were map-matched to OpenStreetMap roadway attributes and supplemented with image-based manual coding to classify functional class, posted speed limit, lane count, centerline configuration, and bicycle infrastructure type. Ordinary least-squares regression models with participant-clustered standard errors were applied to evaluate infrastructure effects and speed-distance relationships, including interactions with driver speeding behavior.
Results
Passing speed varied substantially across roadway contexts, increasing with functional class and posted speed limit. Mean passing speed ranged from approximately 10 m/s (residential roads) to nearly 15 m/s (primary and four-lane facilities). In contrast, lateral clearance varied modestly across contexts, with mean values generally between 2.2 and 3.0 m and substantial overlap among categories. After adjustment for roadway characteristics, standard painted bicycle lanes were associated with reduced passing distance (β = -0.31 m) and higher passing speed (β = +1.36 m/s) relative to roads without bicycle accommodation. Shoulders were associated with the highest passing speeds and no meaningful increase in clearance. The overall compensation slope, defined as the increase in lateral clearance with increasing speed, was small: 0.016 m per 1 m/s (∼2.8 inches per 10 mph). Modest but statistically significant compensation was observed among drivers exceeding the speed limit and on higher-speed, higher-capacity roadways.
Conclusions
Roadway context strongly influenced passing speed, whereas lateral clearance adjustments were comparatively small and inconsistent. Across various naturalistic environments, lateral clearance did not increase proportionally at higher speeds. On-road bicycle infrastructure without accompanying speed management did not reliably reduce high-speed passing and, in some contexts, was associated with closer overtaking. These findings demonstrate the need for vehicle speed reduction and improved separation of bicycles from motor vehicle traffic to reduce crash injury risk.