Replay : anÃlise e depuraÃÃo de sistemas multiagente baseado em conhecimento lÃgico orientado a objetos, com aplicaÃÃo a Robocup

AUTOR(ES)
DATA DE PUBLICAÇÃO

2005

RESUMO

In recent years, many different methodologies and tools for Agent-Oriented Software Engineering have been proposed. Yet, almost all of them deal only with a restricted sub-class of Multi-Agent Systems (MAS): those composed of agents whose sensors and effectors are limited to the exchange of messages with other agents, and those whose reasoning is purely reactive and implemented using domain-specific procedures. In contrast, the MaracatuRFC project aims to apply Software Engineering and Artificial Intelligence techniques to build MAS composed of agents that are authentically situated in a simulated or physical environment.Additionally, the agents can directly sense and orthogonally act upon to any message exchange activity with other agents, and reason in a variety of ways, reactively or deliberatively, using a generic inference engine together with a declarative domain-specific knowledge base. The thesis tackles the issues raised by providing, for such MAS, an integrated set of service that covers both debugging the encapsulated reasoning of each agent and analyzing the emerging cooperative behavior of agent teams. Furthermore, it proposes a set of solutions for these issues and validates them through the development of the Replay tool that provides such services for the case study of RoboCup simulated robotic soccer team engineering

ASSUNTO(S)

logic programming sistemas multiagente multi-agent systems behaviour analisys and debugging robot soccer ciencia da computacao futebol de robà anÃlise e depuraÃÃo de comportamento programaÃÃo lÃgica

ACESSO AO ARTIGO

Documentos Relacionados