Authors: Mohammed Chennoufi; Fatima Bendella; Maroua Bouzid
Addresses: Faculté des Mathématiques et Informatique, Département d'Informatique, Université des Sciences et de la Technologie d'Oran, Mohamed Boudiaf USTO, BP 1505, Oran 31000, Algeria ' Faculté des Mathématiques et Informatique, Département d'Informatique, Université des Sciences et de la Technologie d'Oran, Mohamed Boudiaf USTO, BP 1505, Oran 31000, Algeria ' GREYC, Campus 2, Sciences 3, Basse-Normandie, BP 5186, Caen 14032, France
Abstract: This paper proposes a new approach for multi-agent planning and decision support. The conventional algorithms such as Dijkstra, A* cannot solve complex problems with spatio-temporal constraints. So we are interested in developing a new strategy for the best path based on BDI agents for an emergency evacuation problem of a population crowd, besides the study of the macroscopic behaviour emerging from simple interactions between agents by decreasing the evacuation time which is a challenge and a very complex task. Multi-agent systems are well suited to modelling such systems. The idea is to make a two-dimensional modelling of the environment as a quadtree graph and an hybrid architecture: A* search from the node, where the individual is located to direct it to the best exit node while adding physiological factors to this research, a robust method for collision avoidance and decision support to help the agent will replace the initial destination with anew one. Our model is implemented and tested with java and Netlogo 5.2.1 platform.
Keywords: complex system; A*; multi-agent systems; MAS; crowd; path; decision support; planning; evacuation; simulation; emergence.
International Journal of Operational Research, 2020 Vol.38 No.3, pp.343 - 363
Received: 23 Nov 2016
Accepted: 09 Oct 2017
Published online: 05 May 2020 *