INCORPORATING PLANNING INTO BDI SYSTEMS
Felipe Rech Meneguzzi, Avelino F. Zorzo, Michael Da, Costa Mora, Michael Luck · Research Portal (King's College London) · 2007
Abstra t. Many ar hite tures of autonomous agent have been proposed throughout AI resear h. The most ommon ar hite-tures, BDI, are pro edural in that they do no planning, seriously urtailing an agent's ability to ope with unforeseen events. In this paper, we explore the relationship between propositional planning systems and the pro ess of means-ends reasoning used by BDI agents and dene a mapping from BDI mental states to propositional planning problems and from propositional plans ba k to mental states. In order to test the viability of su h a mapping, we have implemented it in an extension of a BDI agent model through the use of Graphplan as the propositional planning algorithm. The implemented prototype was applied to model a ase study of an agent ontrolled produ tion ell. Key words. Propositional Planning, Agent Models and Ar hite tures, BDI, X-BDI 1. Introdu tion. Development of autonomous rational agents has been one of the main drivers of arti ial intelligen e resear h for some time [37℄. Initial eorts fo used on disembodied means-ends reasoning with the development of problem-solving systems and generi planning systems, su h as STRIPS [15℄, later evolving into the idea of embodied problem solving entities (i.e. agents) [37℄. In this line of resear h, one of the most widely studied models of autonomous agents has been that supported by the mental states of beliefs, desires and intentions [7℄, or the BDI model. While eorts towards dening BDI ar hite tures have been sustained and signi ant, resulting in both theoreti al [34 ℄ and pra ti al ar hite tures [14℄, they have also led to a dis onne t