Talk Anastassia Vybornova at CRBAM21
October 15 @ 10:00 - 10:15
Network algorithms for the identification and classification of gaps in urban bicycle networks based on OSM data
What is the best location to build new bicycle infrastructure in a city? This planning decision can be conceptualized as an optimization problem: the goal is to find the most efficient solution – that is, the one that has the highest (positive) impact on the bicycle network at least cost. However, identifying locations that will significantly improve network quality is far from being a straightforward task. The first challenge is to put a number on network quality improvement in order to make different planning decisions comparable; the second is the often cumbersome collection and processing of data required by the chosen approach (e.g. input from large-scale user surveys).
Our data-driven, computational approach for the identification of gaps in bicycle networks simultaneously addresses both these challenges. We present an algorithm that identifies and rates (by relevance) gaps in a bicycle network, based solely on the input of topological features which are readily available as open-source data from OSM (OpenStreetMap). For this purpose, we define “gaps” as segments (of flexible maximum length) of existing streets which, if provided with a bicycle facility, improve the overall quality of the network. We assess the quality of the network based on network connectivity measures from graph theory, in particular edge betweenness centrality. We furthermore assume that cyclists prefer to use protected bicycle infrastructure and are willing to take a certain detour to maximize the percentage of route spent on bicycle facilities.
To showcase the applicability of our approach, we present the gaps identified by the algorithm in the bicycle network of Copenhagen, classified and ranked by relevance (quality impact). We compare our results with the city’s current bicycle network development plan (Cykelsti Prioriteringsplan 2017-2025) to assess the validity of our findings. Our work shows how network analysis based on open-source topological data can serve as a powerful and cost-efficient tool for decision-making support in bicycle network planning.