1. How do we go with the flow? - 4CBLW00-01

Measuring, understanding, predicting and managing the behavior of large crowds in public spaces is a societally critical endeavor and a deep scientific challenge at the interface between physics, mathematical modeling, AI, big-data, computer vision, but also social psychology.  Crowds flow as pedestrians interact mutually as well as with the environment generating extremely complex dynamics with broad stochasticity. In recent years, we acquired the capacity of measuring with very high accuracy crowd dynamics: this means tracking anonymously and individually each pedestrian within hundreds of square meters and with 24/7 schedules. This allows us to create datasets with hundreds of thousands of trajectories: the perfect foundations for modeling, phenomenological understanding, but also environment/geometry optimization and control.   The students will get access to the datasets of the crowdflow research groups that include tracking within train stations, large events such as concerts and festivals, and indoor environments (dataset size O(10⁴-10⁷) trajectories).  On these bases, students can formulate their challenge that can be oriented both to fundamental understanding in relation to crowd flows in built environment (modeling of observed dynamics through (stochastic/differential) equations, probabilistic models, phenomenological analyses, validation of hypothesis), data-driven analytics (big-data analysis, pattern discovery, data-driven modeling, optimal data selection/active learning), but also technological application (large-scale simulation, facility design improvement, optimal control). Thus, this challenge can be phrased both as an Open research challenge or as an Engineering design challenge.