Introduction to the Special Issue on Knowledge Engineering Tools and Techniques for Automated Planning and Scheduling Systems

摘要

Competitions are frequently fostering and speeding-up development in a given area. Since 1998, the biannual International Planning Competition (IPC) has been organized with the goal to promote planning technology. Indeed, this competition is very successful — it led to rapid development of planning techniques, established a common language for communicating the dynamics of domain models (PDDL) and helped in the validation and scrutiny of planning algorithms. However, the scope of IPC is limited to planning algorithms while omitting the other aspects of solving planning problems, for example design of domain models and gathering application knowledge which is crucial for efficiency of planning systems. Consequently, the contribution of IPC to solving real-life problems is limited. No matter how efficient or powerful Planning and Scheduling (P&S) engines are, they are only as good as the application knowledge that they use. If the P&S domain model is flawed, the resulting P&S application will be flawed. These ideas stood behind the new International Competition on Knowledge Engineering for Planning and Scheduling (ICKEPS). The first ICKEPS was collocated with the ICAPS 2005 conference in Monterey, U.S.A. and seven systems participated in the competition. Their authors have been invited to extend the description of their systems into a journal form and submit to this Special Issue. From these submissions, three papers have been selected to form this special issue on knowledge engineering tools and techniques for automated planning and scheduling. Not coincidentally, two of these selected papers represent the systems that won the first Competition on Knowledge Engineering for Planning and Scheduling, namely GIPO and ARMS.