Decision Support for Crowd Control: Using Genetic Algorithms with Simulation to Learn Control Strategies

In this paper we describe the development of a decision support system for crowd control. Decision support is provided by suggesting a control strategy needed to control a specific current riot situation. Such control strategies consists of deployment of several police barriers with specific barrier positions and barrier strengths needed to control the riot. The optimal control strategy for the current situation is found by comparing the current situation with pre-stored example situations of different sizes. The control strategies are derived for these pre-stored example situations by using genetic algorithms where successive trial strategies are evaluated using stochastic agent-based simulation.