Planification multi-agent et diagnostic stratégique
نویسندگان
چکیده
When several agents act in a common environment their plans may interfere. The predicted outcome of each plan may be altered and the usual notion of valid plan of classical (mono-agent) planification is not sufficient. In this paper we extend this framework to the multi-agent case. We show how the best plans of rational agents can be characterized using game-theoretic notions, especially Nash equilibrium. We also identify the scenarios for which a cooperation between agents is likely to occur and show that many strategic information can be derived from the representation of the interaction.
منابع مشابه
Résolution coopérative de problèmes intégrant planification et interaction
This article deals with a human model of cooperative problem solving based on a psychological experiment. Our main goal is to improve the design of open multi-agent systems able to interact with humans. We aim at analyzing, modelling and simulating human capabilities of planning and interacting in a cooperative problem solving context. This paper goes on to present an original architecture call...
متن کاملUtilisation de croyances heuristiques pour la planification multi-agent dans le cadre des Dec-POMDP. (Using heuristic belief points for Dec-POMDP planning)
متن کامل
Un modèle de composition automatique et distribuée de services web par planification
Web services advent as an inevitable technology of the Web and its dissimination on a large scale, poses the problem of their automatic composition. Indeed, one of the most important obstacle to the development of web services oriented architectures relies on the manual generation of composite services by human experts. In order to overtake this approach, we propose in this article a novel arch...
متن کاملProgrammation dynamique à base de points pour la résolution des DEC-POMDPs
RÉSUMÉ. Nous présentons un nouvel algorithme de planification pour la construction de systèmes multi-agents réactifs et situés pouvant se modéliser par des processus de décision de Markov décentralisés (DEC-POMDP). Cet algorithme est fondé sur la programmation dynamique à base de points. Il est dérivé de techniques de programmation dynamique optimale utilisées pour résoudre des jeux stochastiqu...
متن کاملLa géo-simulation multi-agent. Un support pour la planification dans un environnement réel
In this paper, we propose an agent-based geo-simulation approach to assist planners when planning under strong spatial constraints in a real space. The approach consists in drawing a parallel between the real environment (for example, a forest on fire) and the simulated environment based on georeferenced data. This virtual environment uses software agents which are aware of the space and equipp...
متن کامل