Intelligent Agent Simulator in Massive Crowd

Abstract

Crowd simulations have many benefits over real-life research such as in computer games, architecture and entertainment. One of the key elements in this study is to include elements of decision-making into the crowd. The aim of this simulator is to simulate the features of an intelligent agent to escape from crowded environments especially in one-way corridor, two-way corridor and four-way intersection. The addition of the graphical user interface enables intuitive and fast handling in all settings and features of the Intelligent Agent Simulator and allows convenient research in the field of intelligent behaviour in massive crowd. This paper describes the development of a simulator by using the Open Graphics Library (OpenGL), starting from the production of training data, the simulation process, until the simulation results. The Social Force Model (SFM) is used to generate the motion of agents and the Support Vector Machine (SVM) is used to predict the next step for intelligent agent.