Smart Crowd Management: The Data, the Users and the Solution

Abstract

This research project is situated in the domain of smart crowd management, a domain that is gaining importance because of the challenges that arise from urbanization, but also the opportunities that come with smart cities. While our cities become more crowded every day, they also become smarter, for example by employing pedestrian tracking sensors. However, the datasets that are generated by these sensors do not allow smart crowd management yet, because they are sparse and not linked to the perception of the crowd. This research will tackle these issues in three steps. First, pedestrian counts will be estimated on streets that have no tracking data by use of deep learning and space syntax data. Next, the perception of crowdedness within the crowd will be linked to the objective pedestrian counts by conducting two user studies, and finally, the resulting subjective pedestrian counts will be used as weights for a routing algorithm. The last step has already been developed as a proof of concept. The routing algorithm, that uses partly simulated data and partly real-time tracking data, has been embedded in a webtool to show stakeholders the potential and goal of this innovative project.