Crowd evacuation analysis using agent-based simulation

• Main Author: Seah, Kwang Beng
• Other Authors: Cai Wentong
• Format: Final Year Project
• Language: English
• Published: 2017
• Subjects: DRNTU::Engineering::Computer science and engineering
• Online Access: http://hdl.handle.net/10356/72819
• Institution: Nanyang Technological University

Crowd simulation in the area of artificial intelligence has seen the rise in importance over the years. The application of tools can be used to simulate different scenarios from crowds in a particular place to evacuation scenario during emergencies. Advancement of agent models regarding complexity allows for more details to be taken into consideration and each simulation to be more realistic. In reality, it is difficult to simulate these scenarios as set up is relatively difficult and possibly dangerous. Crowd simulation provides this platform for the research in how people behave under different conditions without having actually to implement it. In the first part of this project, it requires the understanding of CrowdTool which is developed by a research team in NTU. The uses of CrowdTool varies from simulating crowds in an enclosed room, evacuation from a building, riot and larger scale emergency evacuations. We make use of CrowdTool to carry out a systematic inquiry for crowd simulation. Understanding the crowd and representing it using computational modelling. CrowdTool allows behaviours to be represented using cognitive models and it simulates the human-like behaviour in different scenarios. The underlying works of CrowdTool entails an understanding of complex human behaviours such as senses, situational consciousness, choice making and actions itself. The actions including movement speed, collision avoidance, direction etc. In this project, we make use of CrowdTool to generate scenarios that resemble an evacuation and analyse the evacuation time concerning various scenario. The analysis data obtained suggest the use of intelligent signboards shortens the evacuation time of agents and it can be applied to the real-world situation where more lives can be saved in the time of an emergency.