| | |
Summary: Learning Plan Rewriting Rules
Jos´e Luis Ambite, Craig A. Knoblock & Steven Minton
Information Sciences Institute
University of Southern California
4676 Admiralty Way, Marina del Rey, CA 90292, USA
{ambite, knoblock, minton}@isi.edu
Abstract
Planning by Rewriting (PbR) is a new paradigm for ef-
ficient high-quality planning that exploits plan rewrit-
ing rules and efficient local search techniques to trans-
form an easy-to-generate, but possibly suboptimal,
initial plan into a high-quality plan. Despite the ad-
vantages of PbR in terms of scalability, plan qual-
ity, and anytime behavior, PbR requires the user to
define a set of domain-specific plan rewriting rules
which can be difficult and time-consuming. This pa-
per presents an approach to automatically learning the
plan rewriting rules based on comparing initial and op-
timal plans. We report results for several planning do-
mains showing that the learned rules are competitive
|
