Introducing planning in discrete event systems

Abstract

Precedent studies have revealed a strong relationship between discrete events formalisms and formalisms for temporal reasoning in Artificial Intelligence. Both areas typically deal with temporal or dynamic domains, but under different points of view. Discrete events formalisms are intended for the analysis of the represented system, predicting its behavior by simulation. Artificial Intelligence approaches pay special attention to the adequacy of the representation to be used, which usually allows solving also planning problems. This paper is a study on the introduction of planning techniques into the DEVS model, used for modeling of discrete event systems. To this aim, we have chosen a basic representation, the Generalized Magnitudes temporal reasoning formalism, which let us represent a complete DEVS model of a system and simultaneously provides interesting properties for planning. The paper identifies the basic steps that a planning. algorithm should carry out providing also some examples. Finally, it poses possible problems that may appear and establishes the future research lines.

This research was supported in part by the Government of Galicia (Spain), grant XUGA10503B/96