A General Approach for Configuring PDDL Problem Models


  • Mauro Vallati University of Huddersfield
  • Ivan Serina University of Brescia




Artificial Intelligence, Automated Planning, Knowledge Configuration


The development of a large number of domain-independentplanners is leading to the use of planning engines in a widerange of applications. This is despite the complexity issues inherent in plan generation, which are exacerbated by the separation of planner logic from domain knowledge. However, this separation supports the use of reformulation and configuration techniques, which transform the model representation in order to improve the planner's performance. In this paper, we investigate how the performance of domain-independent planners can be improved by problem model configuration. We introduce a fully automated method for this configuration task, that considers problem-specific aspects extracted by exploiting a problem- and domain-independent representation of the instance. Our extensive experimental analysis shows that this reformulation technique can have a significant impact on planners' performance.




How to Cite

Vallati, M., & Serina, I. (2018). A General Approach for Configuring PDDL Problem Models. Proceedings of the International Conference on Automated Planning and Scheduling, 28(1), 431-435. https://doi.org/10.1609/icaps.v28i1.13869