Crowd Simulator using Genetic Algorithm

Crowd management is a problem that needs to be solved in an optimized way. Several algorithms exist which save time to reach the destinations. This paper proposes the Genetic Algorithm as an alternative to designing a conventional controller via algorithmic means. The paper aims to showcase the proposal of a type of crowd simulator by the use of genetic algorithms to select for a simulator with desirable properties, such as low collision and staying on the path. The crowd simulator is a waypoint-based implementation that navigates a constantly populating crowd to their destination in an urban setting, described as a potential field. The crowd is steered by a common controller, which instructs each individual based on its genetic code. The genetic code is measured for Fitness based on the total collisions and the time spent across the potential field.