The Effect of Domain Modeling on Efficiency of Planning

It is known that a planning domain model influences performance of planners, but so far there have been only limited studies how exactly models influence planners’ performance. There are also no guidelines how to design efficient planning models. This paper studies several modeling techniques, namely factored and structured representations of states, heuristics, and control knowledge and their influence on performance of planning algorithms when iterative deepening and branchand-bound search procedures are used. The Picat planning module is used to demonstrate the techniques on classical planning domains from International Planning