Traffic conflicts of speed pedelecs and bicycles: A naturalistic driving study
DOI: 10.1016/j.aap.2021.106201
archive: archived pipeline: cataloged verified
Abstract
This may be the author’s version of a work that was submitted/accepted for publication in the following source: Vlakveld, Willem, Mons, Celina, Kamphuis, Kas, Stelling, Agnieszka, & Twisk, Divera (2021) Traffic conflicts involving speed-pedelecs (fast electric bicycles): A naturalistic riding study. Accident Analysis and Prevention, 158, Article number: 106201. This file was downloaded from: https://eprints.qut.edu.au/213610/ © 2021 Elsevier Ltd This work is covered by copyright. Unless the doc
Summary
Naturalistic riding study (Vlakveld, Mons, Kamphuis, Stelling, Twisk; SWOV + QUT CARRS-Q; Accident Analysis & Prevention 158:106201) using forward- and backward-facing cameras on speed-pedelecs (electric bicycles with pedal-assist up to 45 km/h) ridden by 28 participants in daily Dutch traffic for 2–3 consecutive weeks. Across 227 hours and 6584 km, 115 traffic conflicts (114 near-crashes with evasive action + 1 minor crash) were identified and analysed via case-cohort comparison against random control moments from the same riders.
Key finding
Conflict partners were predominantly bicycles (51%), then cars/vans (28%) and pedestrians (12%); conflicts most often occurred at intersections and during overtakes of bicycles. Conflict odds were dramatically elevated when bicycles were proximate (OR=43.28) or cars proximate (OR=22.43), when overtaking other users (OR=17.25), at/near intersections (OR=3.94), and on bicycle facilities (OR=1.81), suggesting riding among slower bicycles on cycle paths is the main risk amplifier even though motorised-roadway crashes would be more severe.
Methodology
Naturalistic riding study: 28 participants rode instrumented speed-pedelecs with forward + backward cameras for 2–3 weeks each in everyday Dutch traffic. 227 h / 6584 km of video coded for traffic conflicts (near-crashes + minor crashes); case-cohort logistic-regression analysis compared conflict moments to randomly sampled control moments from the same rider, yielding odds ratios with 95% CIs for proximate users, overtaking, infrastructure type, and intersection proximity.
Sample size: N=28 riders; 227 h video / 6584 km / 115 conflicts
Quality score: 5 / 5