3D interactive visualization of crowd simulations at urban scale

Abstract

Digital representations of the urban environment have constantly increased their complexity: starting from fixed chunks of 2D segment depicting the building base to complex data-sets comprising several layer of information such as multi-resolution GIS data combined with on field 3D data acquisition. Cities in particular and urban development in general emerge from the bottom up, so crowd is an essential part of the city environment but it is often absent from most of the interactive system used to present urban data and landscapes. This paper presents a modular, real-time system based on the open-source framework OpenSceneGraph, providing a realistic and efficient 3D visualization in real-time of complex animated pedestrians within large urban environments, along with tools to visualize, explore and study macroscopic and emerging properties. The system allows to represent large crowd simulations or automatized tracking of real pedestrians streamed from external applications: the component-based design aims to provide efficient workflow for a multi-disciplinary team, including 3D modelers, urban planners, computer scientists, experts of crowd dynamics and other professionals. To evaluate quality, performance and interaction, the system was applied to the case study of “T-Days” in Bologna historical center, using pre-existing 3D open-data assets: some results in terms of performance and workflow among different professionals will be presented.