A Novel Technique for Avoiding Plateaus of Greedy Best-First Search in Satisficing Planning

Abstract

Greedy best-first search (GBFS) is a popular and effective algorithm in satisficing planning and is incorporated into high-performance planners. GBFS in planning decides its search direction with automatically generated heuristic functions. However, if the heuristic functions evaluate nodes inaccurately, GBFS may be misled into a valueless search direction, thus resulting in performance degradation. This paper presents a simple but effective algorithm considering a diversity of search directions to avoid the errors of heuristic information. Experimental results in solving a variety of planning problems show that our approach is successful.

Cite

Text

Imai and Kishimoto. "A Novel Technique for Avoiding Plateaus of Greedy Best-First Search in Satisficing Planning." AAAI Conference on Artificial Intelligence, 2011. doi:10.1609/AAAI.V25I1.8004

Markdown

[Imai and Kishimoto. "A Novel Technique for Avoiding Plateaus of Greedy Best-First Search in Satisficing Planning." AAAI Conference on Artificial Intelligence, 2011.](https://mlanthology.org/aaai/2011/imai2011aaai-novel/) doi:10.1609/AAAI.V25I1.8004

BibTeX

@inproceedings{imai2011aaai-novel,
  title     = {{A Novel Technique for Avoiding Plateaus of Greedy Best-First Search in Satisficing Planning}},
  author    = {Imai, Tatsuya and Kishimoto, Akihiro},
  booktitle = {AAAI Conference on Artificial Intelligence},
  year      = {2011},
  pages     = {985-991},
  doi       = {10.1609/AAAI.V25I1.8004},
  url       = {https://mlanthology.org/aaai/2011/imai2011aaai-novel/}
}