Crowdsourcing Urban Bicycle Level of Service Measures

Crowdsourcing Urban Bicycle Level of Service Measures

The number of cyclists in the US has been growing steadily for the past 20 years. As a result, cities have created bike lanes and bike-shared systems and supported bike to work programs in the past years. However, as bicyclists take the streets, safety is increasingly becoming an important concern. Local governments and bicycle associations are looking into ways of making cycling in urban areas both more attractive and safer. However, one of the main obstacles to decrease the number of bicycle crashes is the lack of information regarding cycling safety at the street level. Bicycle Level of Service (BLOS) models are typically used by local Departments of Transportation (DOTs) to measure bicycle safety. However, these models require extensive information characterizing roadways, which is rarely available for smaller streets or in small towns. This proposal explores how to compute accurate, citywide safety levels based on people's complaints about roads and traffic conditions collected through various crowdsourced platforms. The underlying assumption is that citizens’ complaints and concerns can be used to approximate street safety levels i.e., areas with higher volumes of complaints should be associated to lower levels of service and vice versa. Using data mining and machine learning techniques, this proposal will explore how to automatically compute BLOS values from citizens’ complaints and how to understand the traffic-related reasons behind such safety values. If successful, the expected outcomes will be: (a) an accurate and interpretable model to estimate urban BLOS from user-generated crowdsourced data; (b) a set of easy-to-interpret, actionable items for local DOTs to improve cycling experiences; and (c) a dataset with cycling-related complaints, cycling videos and BLOS measures per road segments for other researchers to advance the state of the art in data-driven cycling safety.

Duration: 
July 2016 - June 2018
Funder: 
National Science Foundation
Total Award Amount: 
$200,000

Principal Investigator: