The goal of our project is to build data-centric models that can be used to simulate interactions involving vehicle passengers, drivers and pedestrians, to assess safety related issues. As a concrete deliverable, we aim at the development of a tool to inform citizens about safety issues in different sites within a urban setting.
Urban mobility involves the management of pedestrian safety regarding traffic accidents. We focus on this issue adopting a data-centric approach based on data from three main sources: (1) video images captured specifically to collect data from pedestrians, (2) information provided by authorities, and (3) existing models for pedestrian behaviour. Our work focuses on i) the combination and treatment of data from these three sources, ii) synthesis of analytical models to extract meaningful patterns from data, and iii) actual generation of relevant information from these data. As a prototype, we plan to build a system to inform citizens about unsafe spots in a city. We believe that the system can be of interest on its own; furthermore the techniques and methodologies developed to design this system can be generalised to a variety of uses, scenarios and domains.