An investigation into the expressive power of PDDL2.1

The planning domain language PDDL2.1, used in the 3rd International Planning Competition, has sparked off some controversy in the planning community as researchers consider its expressive power and the ease with which interesting domain models can be constructed in the language. In this paper we show that the expressive power of PDDL2.1 is much greater than is commonly believed. We demonstrate that PDDL2.1 can model many of the domain features often claimed to lie beyond its modelling capability. In so doing we provide a means by which powerful domain features and language constructs can be given a semantics in terms of the canonical basis of PDDL2.1