Parallelizing State Space Plans Online

Searching for parallel solutions in state space planners is a challenging problem, because it would require the planners to branch on all possible subsets of parallel actions, exponentially increasing their branching factor. We introduce a variant of our heuristic state search planner AltAlt, which generates parallel plans by using greedy online parallelization of partial plans. Empirical results show that our online approach outperforms post-processing (offline) techniques in terms of the quality of the solutions returned.

[1]  Bart Selman,et al.  Pushing the Envelope: Planning, Propositional Logic and Stochastic Search , 1996, AAAI/IAAI, Vol. 2.

[2]  Fernando Pereira,et al.  Non-Projective Dependency Parsing using Spanning Tree Algorithms , 2005, HLT.

[3]  Avrim Blum,et al.  Fast Planning Through Planning Graph Analysis , 1995, IJCAI.

[4]  Subbarao Kambhampati,et al.  Planning graph as the basis for deriving heuristics for plan synthesis by state space and CSP search , 2002, Artif. Intell..

[5]  Joakim Nivre,et al.  Pseudo-Projective Dependency Parsing , 2005, ACL.

[6]  Eva Hajicová,et al.  Issues of Projectivity in the Prague Dependency Treebank , 2004, Prague Bull. Math. Linguistics.

[7]  Subbarao Kambhampati,et al.  AltAlt: Combining the Advantages of Graphplan and Heuristic State Search , 2000 .

[8]  Subbarao Kambhampati,et al.  Solving Planning-Graph by Compiling It into CSP , 2000, AIPS.

[9]  Keith Hall,et al.  Corrective Modeling for Non-Projective Dependency Parsing , 2005, IWPT.

[10]  Patrik Haslum,et al.  Admissible Heuristics for Optimal Planning , 2000, AIPS.

[11]  Christer Bäckström.  Computational Aspects of Reordering Plans , 1998, J. Artif. Intell. Res..

[12]  Subbarao Kambhampati,et al.  AltAltp: Online Parallelization of Plans with Heuristic State Search , 2003, J. Artif. Intell. Res..

[13]  Victor H. Yngve,et al.  A model and an hypothesis for language structure , 1960 .

[14]  Nils J. Nilsson,et al.  Artificial Intelligence , 1974, IFIP Congress.