Mobile Crowd Steering Evacuation Model via Agent Based Simulation by Applying the Social Force Model

• Main Authors: Norhaida, Binti Hussain, Wai Shiang, Cheah, Wai Loke, Seng
• Format: Thesis
• Language: English; English; English
• Published: Universiti Malaysia Sarawak 2025
• Subjects: Q Science (General)
• Online Access: http://ir.unimas.my/id/eprint/49590/

Mobile crowd evacuation systems are cutting-edge technologies that direct and control huge masses during emergencies including fires, disasters, and terrorism. Mobile technology, usually smartphone apps, provides evacuation victims with rapid support, information, and updates. Many researchers have created simulation evacuation models to test evacuation systems' ability to optimize evacuation time. Simulation models can help policymakers evaluate evacuation plans by assessing their efficacy. In addition, simulated data can overcome the lack of genuine datasets for specific crowd situations. It overcomes the cost of capturing and pre-processing genuine crowd datasets. Famous simulation models include agent-based crowd guiding. The evaluation states that the existing agent-based steering model computes the complete set of collision-free paths with the least travel time among agents to make path decisions based on updated information. Updated information is contextual. Context includes information about an entity's context, such as its type of activity, time, location, agent velocity, and neighbouring agents' positions and velocities. The agent steering evacuation model can replicate fire evacuation, although guiding humans is untested. The current crowd simulation model describes context information, however how to simulate information dissemination in mobile crowd guiding evacuation remains unknown. As mobile crowd directing system devotes considerable attention, without such aspects it's hard to test its effectiveness. This study describes a mobile crowd evacuation simulation that effectively distributes emergency information in disaster scenarios. The agent simulation model covers all human characteristics considered for receiving evacuation messages indoors. The NetLogo-simulated agent crowd steering model includes context management, path planning, role selection, and movement modules. Using the original Social Force Model (SFM), a benchmark result is created to validate the model. The context manager, path planning, and movement module of the original SFM were used to create the simulation model. The communication model is added to the simulation model to demonstrate the use of word-of-mouth during evacuation. A role selection algorithm chooses the leader before the evacuation. To simulate individuals following a leader during evacuation, the willingness parameter is added. Simulation models replicate victim agent and leader-follower behaviour using leader-follower models. Simulations show that grouping tactics and message forwarding increase evacuation time by 4–31%. When there are 200 to 500 evacuees, evacuation takes less time than in Chapter 3. Chapter 5 simulations show that adding the willingness parameter reduced evacuation time. In low-cooperation scenarios, the willingness parameter reduced evacuation time just slightly, suggesting little impact. The study found that evacuation is less effective when communities with weak cooperation have enough evacuation information.