Towards Improving the Comprehension of HTN Planning Domains by Means of Preconditions and Effects of Compound Tasks

Abstract

Hierarchical Task Network (HTN) planning is a paradigm that offers engineers a formalism for modeling planning domains in terms of possible decompositions of compound tasks. A complete decomposition of a compound task results in totally or partially ordered primitive tasks, i.e., plans in the classical sense. Existing specification languages for HTN domains, such as HDDL, do only allow the assertion of preconditions and effects for primitive tasks but not for compound ones. Recently, a method for inferring preconditions and effects for compound tasks was proposed. It was hypothesized that inferred preconditions and effects can aid knowledge engineers in understanding HTN domain specifications and in predicting the meaning of particular compound tasks. We describe preliminary results from a study that supports this hypothesis and discuss future research directions. Show more