A Cloud Based Simulation Service for 3D Crowd Simulations

Abstract

Crowd simulation can play a crucial role when it comes to the design of Smart Environments. Crowd simulation can give insights on the flow of pedestrian in particular facilities and explore the interplay between ambient intelligence deployments and users. Most researchers develop crowd simulations using commercial game engines built with the editors they usually provide. This prevents a deeper experimentation with the problems of crowd simulation and enforces to stick to the development paradigm of the tool. As a consequence, it couples the scientific experimentation that produces the crowd model with the actual construction of the simulation tool. Besides, a crowd simulation may require more resources than those available to the scientist. A solution would be to conceive crowd simulation as a service that, on the one hand, it allowed scientists to experiment with the latest advances without the burden of installing elements or acquiring expensive computational resources; and, on the other hand, it enabled developers to evolve the tool in a scalable way. The contribution of the paper is a framework that enables the "simulation as a service" approach for crowd simulations when they are run with a 3D representation. As a proof of concept, the paper illustrates how crowd simulations can be used to generate datasets that allow studying the deployment of sensors in a large facility.