ENABLING A NEW DATA SCIENCE FOR URBAN ACCESSIBILITY FOR ALL

Abstract

We are developing new data collection approaches that use a combination of remote crowdsourcing, machine learning, and online map imagery. Our newest effort, called Project Sidewalk, enables online crowdworkers to remotely label pedestrian-related accessibility problems by virtually walking through city streets in Google Street View. Aided, in part, by PacTrans funding, we have now deployed Project Sidewalk into ten cities across the world. Overall, 11,000 users have contributed over 720,000 labels and 400,000 validations. To our knowledge, this is the largest, most granular open dataset on sidewalk accessibility in existence. This unprecedented dataset enables new types of urban accessibility analyses not previously possible, which is the focus of our work and our report. Specifically, we report on the (1) expansion of Project Sidewalk into three additional cities, including La Piedad, Mexico, Oradell, New Jersey, and Amsterdam, The Netherlands; (2) an initial correlative analysis of how sidewalk accessibility/condition corresponds to socioeconomic factors; and (3) tool development and an initial study of combining Crowd+AI techniques to study how sidewalk accessibility is changing over time in cities