Directed search for generalized plans using classical planners

Siddharth Srivastava, Neil Immerman, Shlomo Zilberstein, Tianjiao Zhang

Research output: Chapter in Book/Report/Conference proceedingConference contribution

13 Citations (Scopus)

Abstract

We consider the problem of finding generalized plans for situations where the number of objects may be unknown and unbounded during planning. The input is a domain specification, a goal condition, and a class of concrete problem instances or initial states to be solved, expressed in an abstract first-order representation. Starting with an empty generalized plan, our overall approach is to incrementally increase the applicability of the plan by identifying a problem instance that it cannot solve, invoking a classical planner to solve that problem, generalizing the obtained solution and merging it back into the generalized plan. The main contributions of this paper are methods for (a) generating and solving small problem instances not yet covered by an existing generalized plan, (b) translating between concrete classical plans and abstract plan representations, and (c) extending partial generalized plans and increasing their applicability. We analyze the theoretical properties of these methods, prove their correctness, and illustrate experimentally their scalability. The resulting hybrid approach shows that solving only a few, small, classical planning problems can be sufficient to produce a generalized plan that applies to infinitely many problems with unknown numbers of objects.

Original languageEnglish (US)
Title of host publicationICAPS 2011 - Proceedings of the 21st International Conference on Automated Planning and Scheduling
Pages226-233
Number of pages8
StatePublished - Oct 27 2011
Externally publishedYes
Event21st International Conference on Automated Planning and Scheduling, ICAPS 2011 - Freiburg, Germany
Duration: Jun 11 2011Jun 16 2011

Other

Other21st International Conference on Automated Planning and Scheduling, ICAPS 2011
CountryGermany
CityFreiburg
Period6/11/116/16/11

Fingerprint

Planning
Directed search
Hybrid approach
Merging
Scalability

ASJC Scopus subject areas

  • Information Systems and Management

Cite this

Srivastava, S., Immerman, N., Zilberstein, S., & Zhang, T. (2011). Directed search for generalized plans using classical planners. In ICAPS 2011 - Proceedings of the 21st International Conference on Automated Planning and Scheduling (pp. 226-233)

Directed search for generalized plans using classical planners. / Srivastava, Siddharth; Immerman, Neil; Zilberstein, Shlomo; Zhang, Tianjiao.

ICAPS 2011 - Proceedings of the 21st International Conference on Automated Planning and Scheduling. 2011. p. 226-233.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Srivastava, S, Immerman, N, Zilberstein, S & Zhang, T 2011, Directed search for generalized plans using classical planners. in ICAPS 2011 - Proceedings of the 21st International Conference on Automated Planning and Scheduling. pp. 226-233, 21st International Conference on Automated Planning and Scheduling, ICAPS 2011, Freiburg, Germany, 6/11/11.
Srivastava S, Immerman N, Zilberstein S, Zhang T. Directed search for generalized plans using classical planners. In ICAPS 2011 - Proceedings of the 21st International Conference on Automated Planning and Scheduling. 2011. p. 226-233
Srivastava, Siddharth ; Immerman, Neil ; Zilberstein, Shlomo ; Zhang, Tianjiao. / Directed search for generalized plans using classical planners. ICAPS 2011 - Proceedings of the 21st International Conference on Automated Planning and Scheduling. 2011. pp. 226-233
@inproceedings{aa20ec1f170d43f789de999cd2daee02,
title = "Directed search for generalized plans using classical planners",
abstract = "We consider the problem of finding generalized plans for situations where the number of objects may be unknown and unbounded during planning. The input is a domain specification, a goal condition, and a class of concrete problem instances or initial states to be solved, expressed in an abstract first-order representation. Starting with an empty generalized plan, our overall approach is to incrementally increase the applicability of the plan by identifying a problem instance that it cannot solve, invoking a classical planner to solve that problem, generalizing the obtained solution and merging it back into the generalized plan. The main contributions of this paper are methods for (a) generating and solving small problem instances not yet covered by an existing generalized plan, (b) translating between concrete classical plans and abstract plan representations, and (c) extending partial generalized plans and increasing their applicability. We analyze the theoretical properties of these methods, prove their correctness, and illustrate experimentally their scalability. The resulting hybrid approach shows that solving only a few, small, classical planning problems can be sufficient to produce a generalized plan that applies to infinitely many problems with unknown numbers of objects.",
author = "Siddharth Srivastava and Neil Immerman and Shlomo Zilberstein and Tianjiao Zhang",
year = "2011",
month = "10",
day = "27",
language = "English (US)",
isbn = "9781577355038",
pages = "226--233",
booktitle = "ICAPS 2011 - Proceedings of the 21st International Conference on Automated Planning and Scheduling",

}

TY - GEN

T1 - Directed search for generalized plans using classical planners

AU - Srivastava, Siddharth

AU - Immerman, Neil

AU - Zilberstein, Shlomo

AU - Zhang, Tianjiao

PY - 2011/10/27

Y1 - 2011/10/27

N2 - We consider the problem of finding generalized plans for situations where the number of objects may be unknown and unbounded during planning. The input is a domain specification, a goal condition, and a class of concrete problem instances or initial states to be solved, expressed in an abstract first-order representation. Starting with an empty generalized plan, our overall approach is to incrementally increase the applicability of the plan by identifying a problem instance that it cannot solve, invoking a classical planner to solve that problem, generalizing the obtained solution and merging it back into the generalized plan. The main contributions of this paper are methods for (a) generating and solving small problem instances not yet covered by an existing generalized plan, (b) translating between concrete classical plans and abstract plan representations, and (c) extending partial generalized plans and increasing their applicability. We analyze the theoretical properties of these methods, prove their correctness, and illustrate experimentally their scalability. The resulting hybrid approach shows that solving only a few, small, classical planning problems can be sufficient to produce a generalized plan that applies to infinitely many problems with unknown numbers of objects.

AB - We consider the problem of finding generalized plans for situations where the number of objects may be unknown and unbounded during planning. The input is a domain specification, a goal condition, and a class of concrete problem instances or initial states to be solved, expressed in an abstract first-order representation. Starting with an empty generalized plan, our overall approach is to incrementally increase the applicability of the plan by identifying a problem instance that it cannot solve, invoking a classical planner to solve that problem, generalizing the obtained solution and merging it back into the generalized plan. The main contributions of this paper are methods for (a) generating and solving small problem instances not yet covered by an existing generalized plan, (b) translating between concrete classical plans and abstract plan representations, and (c) extending partial generalized plans and increasing their applicability. We analyze the theoretical properties of these methods, prove their correctness, and illustrate experimentally their scalability. The resulting hybrid approach shows that solving only a few, small, classical planning problems can be sufficient to produce a generalized plan that applies to infinitely many problems with unknown numbers of objects.

UR - http://www.scopus.com/inward/record.url?scp=80054834173&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=80054834173&partnerID=8YFLogxK

M3 - Conference contribution

SN - 9781577355038

SP - 226

EP - 233

BT - ICAPS 2011 - Proceedings of the 21st International Conference on Automated Planning and Scheduling

ER -