Executive Summary
This deliverable describes the efforts done during periods 1 and 2 in the Work Package 2 of the CrowdDNA project towards developing a new crowd simulator algorithm tailored to model both macro and micro-level crowd characteristics. As a reminder, the overall objective of WP2 is to deliver a new generation of crowd simulation techniques that can predict crowd behaviors at macroscopic scales from numerical models of physical interactions.
To this end, the consortium has worked on several key ingredients for a new crowd simulator that will be used to model macro- and micro-level interactions. In particular, we have developed methods to create 3D synthetic crowds from 2D trajectories and integrated several solutions to model and resolve the contact between individuals. Additionally, we have also improved the macroscopic behavior of the crowd by learning a novel neural social physics model that considers the influence of the environment in the crowd behavior.
Figure 1 depicts the different components of the proposed simulator, next to the CrowdDNA partner who led each of the developments.
All in all, our proposed crowd simulation agglutinates several key properties to improve existing over-simplified solutions for crowd simulation that are unable to model micro-level features. The CrowdDNA consortium will leverage the crowd simulator described in this document to achieve the overall goal of the project: the development of tools for a novel of crowd motion analysis.
