The growing sensitivity of citizens and authorities to the dehumanization of cities is promoting innovative concepts and ideas that aim to increase the role of people in the urban environment, consequently reducing the area allocated to vehicles. The current situation favors a frequent interaction between vehicles and pedestrians. In this scenario, combined with the current high levels of stress, it is not rare that such interactions lead to accidents and run-overs, also affecting particularly sensitive groups: the elderly, children and citizens with disabilities.
In this context, the general objective of the Espacio Persona project has been to design an indicator to assess the safety of pedestrians in urban environments according to their structural, functional and dynamic characteristics. To this end, it was proposed to identify and quantify, through images of public space, the most important characteristics for pedestrian road safety, and add them in a single indicator –with the additional advantage of its potential application on any urban space for which the proper data is available. The methods to meet this challenge combine geometric, visual (urban images), and vehicle flow data, with analysis tools based on Artificial Intelligence (Deep Learning).
The resulting micro-indicator characterizes urban areas (in cells between 100 and 300sqm.) according to their level of safety, taking into account the main causes that lead to pedestrian collisions. The implementation of these micro-indicators was based on public and accessible Big Data through open data portals, focusing on the two cities in Spain with the largest population, Madrid and Barcelona. In particular, from the Ayuntamiento de Madrid, the Institut Cartogràfic i Geològic de Catalunya (ICGC), Google Street View, Mapillary, Openstreetmap, the Policía Municipal de Madrid and the Guàrdia Urbana de Barcelona.
As a result of all this, four notable results were achieved: (1) an architecture for the automatic prediction of vehicle-pedestrian accidents based on urban images; (2) an algorithmic process to extract geographic information on the walkable space of cities from public topographic databases; (3) an innovative method for measuring the presence of perceptual impairments in urban environments, mixing urban images and topographical data; and finally (4) a platform for viewing and querying the data produced.
The main and secondary results of the Espacio Persona project are of possible interest to public entities (DGT, municipalities, cartographic institutes, security forces, etc.), private entities (traffic and mobility consultancies, GIS companies, civil works, etc.) and other organizations (parents’ associations, groups with functional diversity, etc.).