Abstrakt
Planning as programming is an approach to automated planning, where the planning domain model is expressed as a program in some (declarative) program- ming language. Then the modeler can exploit all features of that language to encode control knowledge im- portant for efficient planning.
In this paper we study these features in the logic programming language Picat and its planner module. In particular, we use two planning benchmarks, Nomystery and Childsnack, to compare factored and structured representations of states extended by encodings of control knowledge.