00:00/00:00 </>

Espacio persona: Big data to make urban streets safer

Full Text
While the idea of a city built for people is gaining more and more acceptance today, many of our urban environments remain focused and built around the car. Currently, day-to-day life in the city means the frequent interaction between pedestrians and drivers, a situation which can be dangerous, or, at worst, deadly. This project, a collaboration between the Complex Systems group at IN3 (CoSIN3) of the Universitat Oberta de Catalunya (UOC), the Dirección General de Tráfico (DGT) and the Guàrdia Urbana de Barcelona, aims to quantify the issue of car-pedestrian collisions by characterising specific street areas with an indicator of pedestrian safety based on the structural properties of the street. Concretely, the project will generate this safety index for Spain’s two largest cities, Madrid and Barcelona, but the methodology and pipeline are applicable theoretically to any urban setting. As a base unit for measuring pedestrian safety over space, the total pedestrian area of the city (sidewalks and crossings) will be tessellated into small, regular segments. Each of these segments will be assigned various indicator values, from simple geometric properties (distance to the closest pedestrian crossing; width of sidewalk) to more complex measures such as driver visibility. Geometric operations to arrive at these values are performed on a PostGIS-build geo-database, over a variety of data, including street, sidewalk and block geometries, from diverse sources of open GIS data (Instituto Geográfico Nacional, Institut Cartogràfic i Geològic de Catalunya, OpenStreetMap, municipal data sources).Visibility values will be derived from a combination of deep learning technologies with GIS. A deep learning architecture will deliver computer-segmented street-scene images from Google Streetview. Each labeled image will be paired with an image from a simplified 3-dimensional model of the city, replicating its point of view (rendered with open-source 3D mapping software). The model will be clean of all street features (parked cars, trees, etc.) besides sidewalks and buildings. Comparison between the real and simplified images will thus permit the identification of sidewalk areas invisible to drivers due to visual obstructions. The results of the project will be presented as online “heatmap” visualisations of safety indexes for the focus cities, open to the public for browsing and research. Additionally, a purpose-built API (Application Programming Interface) will provide public and private organisations working in the area of traffic safety access to the results for integration in their own internal or public applications ​
This document is licensed under a Creative Commons:Attribution – Non commercial – Share alike (by-nc-sa) Creative Commons by-nc-sa4.0