الفهرس | Only 14 pages are availabe for public view |
Abstract Automated Planning is one of the important applications in Artificial Intelligence that become prominent, recently, in several new technologies and paradigms in a variety of real applications. Planning is a very hard task even in pure search version and looking for optimal solutions can only be hard. In fact, there are few optimal planners and there are a relatively small number of other planners which usually produce good solutions without guaranteeing the optimality. The problem of optimal planning is a hard-combinatorial optimization problem, for which there exist only a few standard algorithmic techniques. Therefore, the main objective of this thesis is to improve the search technique of Hierarchical Task Network (HTN) planning in PANDA planner by adapting Ant system algorithm into refinement planning process by means of plan selection strategy. A few previous methods have been suggested to improve planning solutions to get the optimal plan. Each method has its planner, domain and problem description. The proposed search strategy fuses the Ant system with plan selection strategy to improve optimal plan challenges. The evaluation of our proposed framework showed an outstanding performance against many plan selection strategies. |