HTN-Maker: Learning HTNs with Minimal Additional Knowledge Engineering
Required
This web site maintains the source code and domain files for HTN-Maker, HTN-MakerND, and the (Q-Maker, Q-Reinforce,Q-SHOP) variant (see descriptions below). HTN-Maker was developed under funding by National Science Foundation under Grant No. NSF 0642882. For a broad view of other publications and activities performed under that effort please visit this web site.
- HTN-Maker is an algorithm for learning hierarchical
planning knowledge in the form of decomposition
methods for Hierarchical Task Networks (HTNs). HTN-Maker
takes as input the initial states from a set of classical
planning problems in a planning domain and solutions
to those problems, as well as a set of semantically-annotated
tasks to be accomplished. The algorithm analyzes this semantic
information in order to determine which portions of
the input plans accomplish a particular task and constructs
HTN methods based on those analyses.
Main reference: Hogg, Chad, Munoz-Avila, Hector, and Ugur Kuter. (2008)
HTN-Maker: Learning HTNs with Minimal Additional Knowledge Engineering Required.
Proceedings of the Twenty-Third AAAI Conference on Artificial Intelligence (AAAI-08).
AAAI Press.
PDF
- HTN-MakerND extends HTN-Maker by allowing the learning of methods in nondeterministic planning domains, in which actions may have multiple possible outcomes. The resulting methods can be used by a planner for nondeterministic HTN domains, such as ND-SHOP
Main reference: Chad Hogg, Ugur Kuter, and Hector Munoz-Avila (2009)
Learning Hierarchical Task Networks for Nondeterministic Planning Domains.
Proceedings of the Twenty-first International Joint Conference on Artificial Intelligence (IJCAI-09).
AAAI Press.
PDF
- (Q-Maker, Q-Reinforce,Q-SHOP) We conducted research integrating HTN Learning and Reinforcement Learning. Our formalism associates with HTN methods Q-values that estimate the quality of plans that could be generated using them. This work consists of three algorithms: (1) Q-Maker, which follows a bottom-up procedure starting from input traces to learn HTN methods and initial estimates for their Q-values; (2) Q-Reinforce, which follows a top-down HTN plan generation procedure to refine the Q-value associated with methods; and (3) Q-SHOP, a variant of the HTN planner SHOP that reduces tasks by selecting the appropriate method with highest Q-value
Main reference: Chad Hogg, Ugur Kuter, and Hector Munoz-Avila (2010)
Learning Methods to Generate Good Plans:
Integrating HTN Learning and Reinforcement Learning.
Proceedings of the Twenty-Fourth AAAI Conference on Artificial Intelligence (AAAI-2010).
AAAI Press.
PDF
Downloads:
Note: HTN-Maker and variants are released under terms of the GNU general public license
as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Publications:
Chad Hogg, Ugur Kuter, and Hector Munoz-Avila (2010)
Learning Methods to Generate Good Plans:
Integrating HTN Learning and Reinforcement Learning.
Proceedings of the Twenty-Fourth AAAI Conference on Artificial Intelligence (AAAI-10).
AAAI Press.
PDF
|
Chad Hogg, Ugur Kuter, and Hector Munoz-Avila (2009)
Learning Hierarchical Task Networks for Nondeterministic Planning Domains.
Proceedings of the Twenty-first International Joint Conference on Artificial Intelligence (IJCAI-09).
AAAI Press.
PDF
|
Chad Hogg, Ugur Kuter, and Hector Munoz-Avila (2009)
From Plan Traces to Hierarchical Task Networks
Using Reinforcements: A Preliminary Report.
Proceedings of the IJCAI-09 Workshop on Learning Structural Information from Traces (STRUCK-09).
AAAI Press.
PDF
|
Hogg, Chad, Munoz-Avila, Hector, and Ugur Kuter. (2008)
HTN-Maker: Learning HTNs with Minimal Additional Knowledge Engineering Required.
To appear in Proceedings of the Twenty-Third AAAI Conference on Artificial Intelligence (AAAI-08).
AAAI Press.
PDF
|
Hogg, C. & Munoz-Avila, H. (2007)
Learning of Tasks Models for
HTN Planning.
Proceedings of the ICAPS-07 Workshop on AI Planning and Learning (AIPL). AAAI Press.
PDF
|
Contributors:
Contact about web site: Hector
Munoz-Avila
This material is based upon work supported by the National Science Foundation under Grant No. NSF 0642882. Any opinions, findings and conclusions or recomendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation (NSF)
Last updated: Thur. Apr. 9 11:49:51 EST 2010