Klavins, EricCaspi, AnatBolten, Nicholas2020-08-142020-08-142020-08-142020Bolten_washington_0250E_21932.pdfhttp://hdl.handle.net/1773/45789Thesis (Ph.D.)--University of Washington, 2020Pedestrian informational needs are currently not well-met by existing data infrastructure or public-facing applications, limiting the public's ability to reliably plan trips through public spaces and to understand existing inequities in infrastructure such as network gaps that prevent access to wheelchair users. Modeling pedestrian accessibility requires addressing multiple key problem domains simultaneously, including applying techniques borrowed from the modeling of non-pedestrian transportation networks as well as bringing in outside and new techniques. This dissertation will lay out the numerous problem domains in creating a truly adequate pedestrian network data model, populating that data model, and leveraging it to provide appropriate information to individuals and analysis for use by agencies or the public. The process of building and studying the AccessMap website, an interactive, personalized accessibility map is presented to illustrate the technical and informational barriers that currently stymie work on this issue, including the lack of a standard data schema for pedestrian networks, the need to make use of non-standard data formats, the need to collect fundamental pedestrian network information, the definition of appropriate cost functions for pedestrian mobility, the lack of information regarding pedestrian informational requirements and preferences, and challenges in communicating pedestrian preferences. This work also discusses graph analysis and advocacy enabled by a city-scale pedestrian network and appropriate pedestrian cost functions for graph traversal.application/pdfen-USCC BYAccessibilityGraph theoryMapsOpenStreetMapPedestrian networkWheelchairComputer engineeringArtificial intelligenceComputer scienceElectrical engineeringThesis