Toward Automated Planning Algorithms Applied to Production and Logistics

Abstract

In recent years several studies have been made showing artificial intelligence techniques as enhancement proposals for real practical systems. One of such approaches is automated planning, in which knowledge of the system's behavior, expressed through a model, is used by a piece of software denominated automated planner to infer a sequence of actions capable of bringing the system from some initial state to an objective, a so called plan. To do such, the planner relies on some search algorithm capable of exploring the possibilities exposed by the model, and several different approaches have been used by different planners with varying degrees of success. This paper presents an insight on some of the most consolidated ones, both regarding deterministic and probabilistic domains, and focuses on search techniques and generic heuristics in order to assist the development of new algorithms focused on production and logistics. It also covers the main formal system modeling languages, such as STRIPS, PDDL and PPDDL, used by such planners.