| Nowadays, Intelligent Planning and Planning Recognition are very hot branch in Artificial Intelligence. The main task of Intelligent Planning is find out a serial actions. The serial actions can change the state of world from the initialization to the goal. However, Planning Recognition is used to deduce the goal and the planning that the Agent use, based on observed incomplete actions. In 2000, Ian Miguel build flexible planning, he import the"soft"constraint into Graph planning, this lead the ability of flexible planning have a lot improve in deal with the problems in the real world. On the Other hand, lots of focus is made on the recognition based on the graph planning. Based on two procedures of fast planning, build planning graph and search plan, Hong Jun made the goal method to solve the recognition problem in graph planning problem frame and got success. This method, compare to the other recognition methods, have a great excellence that it need not the warehouse of plans. But goal graph build basis on the theory classical graph plan. Therefore it can not deal with the"soft"constraint action and target.This paper is integrated the method of flexible planning and Goal graph, and research the method of planning recognition. Under the frame of flexible planning, we build the flexible goal graph basis on observed action, initial conditions and goal conditions. Basis on flexible goal graph, we use the causality to prune branches first, use the observed actions to cut more branches. use these two methods will reduce the scale flexible goal graph, and limit the range of plans that Agent can adopt.Flexible planning can get many planning result at same time, it still have many possible results left, when the cut method done. So we discuss the mothod that can recongnize the plan of Agent may be do. The results of experiment shows that the arithmetic is effective in this paper, and can solve the new problem under the frame of flexible planning. |
PDF Full Text Request