A New Pathfinding Evacuation Decision-Making Algorithm in Buildings on Fire Based on Personality Traits
Year: 2024
Type: conference-paper
Abstract: The current research gives a different aspect to pathfinding algorithms in forgetting about an optimum route and offering more value to the current psychological state of a person, which is crucial in not having an ideal simulator where people run where they should be running (optimal route). Still, instead, they move based on their personality traits and reactions. The proposed algorithm is derived from the$\mathrm{A}^{*}$algorithm; it introduces a psychological model known as the OCEAN model or big five for the heterogeneous crowd in a multiagent system that allows decision-making in pathfinding. The OCEAN model has five positive and negative dimensions. To simulate them, we created ten videos in Unity, considering animations with people moving under fire in different scenarios; we used the$\Psi *$algorithm in these simulations to calculate the optimal route. Initially, we evaluated ten different personality trait adjectives. Later, we applied an online survey to validate our model through perceptions of people; the adjectives for pathfinding were reduced to four different types of personality traits due to the limitation that pathfinding can visually offer in a simulator and the relationships that the participants found between said definitions and similar routes thrown by the system. The most relevant results highlight that the$A^{*}$algorithm always yields the optimal evacuation route. That algorithm was modified in$\Psi *$to simulate routes that lose optimality but are close to possible routes that a person with a specific personality type would take. Due to the different reactions of the crowd, our proposal can represent a more approximate real-life scenario of a building on fire.
Institution: Universitat Politècnica de Catalunya
Language: English
Subfield: Ocean Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W4412934125