Hierarchical Task Network Planning as Satisfiability

The satissability paradigm has been hitherto applied to planning with only primitive actions. On the other hand, hierarchical task networks have been successfully used in many real world planning applications. Adapting the satissability paradigm to hierarchical task network planning, we show how the guidance from the task networks can be used to signiicantly reduce the sizes of the propositional encodings. We report promising empirical results on various encodings that demonstrate an orders of magnitude reduction in the solving times.