Concise finite-domain representations for PDDL planning tasks
Artificial Intelligence
| Hi-index | 0.00 |
The generation of the set of all ground actions for a given set of ADL operators, which are allowed to have conditional effects and preconditions that can be represented using arbitrary first-order formulas is a complex process which heavily influences the performance of any planner or pre-planning analysis method. The paper describes a sophisticated instantiation procedure that determines so-called inertia in a given problem representation and uses them to perform simplifications of formulas during the instantiation process. As a result, many inapplicable actions are detected and ruled out from the domain representation yielding a much smaller search space for the planner.