Parallel simulation of a stochastic agent/environment interaction model

Article type: Research Article

Authors: Bouzid, Makram^{a; *} | Chevrier, Vincent^a | Vialle, Stéphane^b | Charpillet, François^a

Affiliations: [a] LORIA UMR 7503, BP 239, 54506 Vandœuvre-lès-Nancy Cedex, France. Tel.: +33 3 83 59 20 51; Fax: +33 3 83 41 30 79; E-mail: [email protected], [email protected], [email protected] | [b] SUPELEC, 2 rue Edouard Belin, 57070 Metz, France. E-mail: [email protected]

Correspondence: [*] Corresponding author.

Abstract: In this article, we address the modeling and simulation of situated agents, together with the unreliability of their sensors and effectors. We propose a model for the simulation of the agent/environment interaction, that integrates uncertainties and errors of agent sensors and effectors. This model is built upon a stochastic modeling of the interaction. We illustrate an application of such a model in the case of simulating a set of mobile autonomous robots evolving in a structured environment (the inside of a building). Moreover, the built simulator has been designed to be executed on parallel computers and we describe its parallel implementation together with major results from a simulation point of view, as well as parallel performances.

DOI: 10.3233/ICA-2001-8302

Journal: Integrated Computer-Aided Engineering, vol. 8, no. 3, pp. 189-203, 2001