Modeling a real-case situation of egress using BDI agents with emotions and social skills

Marion Valette (1, 2) , Benoit Gaudou (3, 4) , Dominique Longin (1) , Patrick Taillandier (2)

1 IRIT-LILaC - Logique, Interaction, Langue et Calcul
2 MIAT INRA - Unité de Mathématiques et Informatique Appliquées de Toulouse
3 IRIT-SMAC - Systèmes Multi-Agents Coopératifs
4 UMMISCO - Unité de modélisation mathématique et informatique des systèmes complexes [Bondy]

To be realistic, evacuation simulations have to consider several aspects of the human psychology that affect their decision-making process. Among them we find social relationships and emotions like fear. The former has been proven to have a great influence on the outcomes of simulations as they modify the behaviour of agents to make them escape in groups. This phenomenon strongly affects the efficiency of the evacuation. The latter impacts the ways the people will try to escape, leading to adaptation and unplanned behaviour. This paper presents an evacuation model that includes cognition with a BDI architecture to represent the way agents do complex reasoning, social relationships and a modelling of fear. The model is applied to simulate the fire of the Rhode Island Station Nightclub in 2003. We shows that after calibration, the model enables to reproduce in a credible way the real event.

Mots clés: BDI; Egress; Simulation; Social relationships; Emotions

Domaines: Base de données [cs.DB]

Marion Valette, Benoit Gaudou, Dominique Longin, Patrick Taillandier. Modeling a real-case situation of egress using BDI agents with emotions and social skills. International Conference on Principles and Practice of Multi-Agent Systems (PRIMA 2018), Oct 2018, Tokyo, Japan. pp.3-18. ⟨hal-02279420v1⟩