Simulation and modelling of crowd control within the context of riots

Abstract: This project aims to shed light on the abilities of simulation to optimize crowd control management strategies. Simulations were chosen because the difficulties of analyzing crowd control management strategies during an ongoing riot. Riots are dangerous to the people involved and experimenting under these conditions could prove costly both in damages to infrastructure and injuries to the people involved. These crowd control management strategies will in the context of this study be the use of altering formation types. More specifically, this study will attempt to answer, "Which crowd-policing agent formation type manages to disperse rioters most optimally?". To answer this research question the game engine Unity was used to create a simulated environment. The crowd-policing agents acted as a static blockade against rioting agents who were instead controlled by a goal-driven boid-algorithm. The goal-driven part refers to boids moving in a specific direction, towards the formation. Three different formation types were used based on US military strategies for crowd control. These formations include the "line"-formation, the "wedge"-formation, and the "checkerboard"-formation. The quantitative measurement used to determine the most optimal formation are the number of separate groups observed during the simulation. If a formation type manages to create more smaller groups, it is seen as more effective due to there being spaces in the rioting crowd. A group was determined to be separate if the agent in that group was not affected by another agents group according to the range of the boid-algorithm. The results indicated that the so called "checkerboard"-formation was most effective in dispersing rioters in the context of this simulation. Hence simulations can be of value to the development of crowd control strategies if used in an appropriate manor. That is if more factors of human interaction during riots are used.