Research And Implementation Of Creating Or Destroying Objects Plan Based On GraphPlan (CDOGP)

Classical planning problems make assumptions about the world as follows: (1) the goals of the planner are a logical description of world states, (2) the actions taken by the planner are the only sources of change in the world, and (3) each action can be described by the conditions under which it can be applied and its effects on the world. These effects are described by a set of facts which are either to be added to or deleted from the world state in which the action is applied. These facts describe the characters of object and state, can not express the creating or destroying object. Graphplan is one of the most popular planners in the 90s of 20 century, and lots of scholars still devote themselves to research under its framework. The main limit of Graphplan is that it only applies to STRIPS problems, especially can not handle the planning problem with creating or destroying objects, but this kind of problem is familiar in real world.This paper makes the elementary study of the Creating or Destroying Objects planning. According to whether the property of the new object is predictable or not, this kind of problem was first divided into two types: planning with predictable object's property and planning with unpredictable object's property. This paper brings forward the idea of transforming objects into propositions, based on which we transform mutex between objects into mutex between propositions. In addition, this paper offers an algorithm called Creating or Destroying Objects in the Graphplan framework (CDOGP) and implements it using C++. We analyze the experimental results, compared with Graphplan the new algorithm not only preserving the highly efficiency of Graphplan, but also can handle Creating or Destroying Objects planning problems that new objects do not interfere with each other.There are a large number of Creating or Destroying Objects planning problems in the real world, so the research in this paper will enlarge the theoretical foundation of intelligent planning and the application domains, it is important for the study of robot control, intelligent game, the comprehension of natural language, and industry produce etc.