Action constraints for planning

Abstract

Recent progress in the applications of propositional planning systems has led to an impressive speed-up of solution time and an in-crease in tractable problem size. In part, this improvement stems from the use of domain-dependent knowledge in form of state constraints. In this paper we introduce a difierent class of constraints: action con-straints. They express domain-dependent knowledge about the use of actions in solution plans and can express strategies which are used by human planners. The use of action constraints results in a tendency to better plans. We explain how to calculate and apply action constraints in the framework of parallel total-order planning, which is the design of the most powerful planners at the moment. We present two classes of action constraints and demonstrate their capabilities in the planner ProbaPla.