Plan Repair in Hybrid Planning

Proceedings of the
31st German Conference on Artificial Intelligence (KI 2008)
Kaiserslautern, Germany
September 23-26, 2008

Hybrid Planning, the integration of hierarchical task decomposition and partial-order-causal-link planning provides a powerful mechanism to solve real-world planning problems. While the generation of hybrid plans is well understood, the problem of how to adequately deal with the various types of execution failures in this context has not yet been addressed.

We present a domain-independent approach to plan repair in hybrid planning. It exploits the generation process of the failed plan by retracting decisions that led to the failed plan fragments. They are selectively replaced by suitable alternatives, and the repaired plan is completed by following the previous generation process as close as possible. This way, a stable solution is obtained, i.e. a repair of the failed plan that causes minimal perturbation. We show how plan repair fits into a formal framework for hybrid planning and present a generic refinement-retraction-and-repair algorithm.

Online Copy

The paper is available on-line as PDF at http://dx.doi.org/10.1007/978-3-540-85845-4_21 via SpringerLink

@InProceedings{bidot:biundo:schattenberg:08,
  author =    {Julien Bidot and Susanne Biundo and Bernd Schattenberg},
  title =     {Plan Repair in Hybrid Planning},
  booktitle = {KI 2008: Advances in Artificial Intelligence, 
               Proceedings of the 31st German Conference on Artificial Intelligence}, 
  Editor =    {Andreas Dengel and Karsten Berns and Thomas Breuel and Frank Bomarius and Thomas R. Roth-Berghofer},
  series=     {Lecture Notes in Artificial Intelligence},
  volume=     {5243},
  pages =     {169-176},
  year =      {2008},
  address =   {Kaiserslautern, Germany},
  month =     september,
  publisher = {Springer}
}