Assessing bicycle safety risks using emerging mobile sensing data

Area of Science:

• Urban Planning
• Transportation Engineering
• Traffic Safety

Background:

• Global rise in electric bicycle (e-bike) ownership strains existing bicycle infrastructure.
• Need to reassess safety risks and infrastructure quality for mixed-traffic environments.
• Lack of precise spatial data for bicycle infrastructure in many regions.

Purpose of the Study:

• To re-evaluate bicycle infrastructure safety risks considering e-bikes and traditional bicycles.
• To introduce a cost-effective mobile sensing method for large-scale bike lane data acquisition.
• To develop and apply a computer vision model for assessing bicycle safety risks.

Main Methods:

• Mobile sensing using bicycles for daytime and nighttime data collection.
• Computer vision-based risk factor assessment model.
• Spatial analysis of collected bicycle lane data in Beijing.

Main Results:

• E-bikes comprised 72.1% of cyclists, with significant rates of non-helmet use (32.3%) and contraflow riding (8.4%).
• Daytime risks include shared/missing dedicated lanes, roadside parking, and poor road conditions.
• Nighttime risks are primarily due to insufficient street lighting.

Conclusions:

• The study highlights the significant impact of e-bikes on cycling safety and infrastructure.
• The proposed mobile sensing methodology is scalable and adaptable for diverse cycling environments.
• Findings provide crucial insights for enhancing bicycle safety policies and urban road design.