Exploring Disjunctive Planning as a Means to Scale-Up Plan Synthesis
Abstract
In this research, we have concentrated on exploring the advantages of disjunctive plan representations in improving planning performance. The results of the research included: (1) development of a system called GP-CSP, which uses direct constraint satisfaction techniques to solve the planning graph of Graphplan; (2) development of a method where the distance heuristics that were originally developed as part of the heuristic state search community can be applied to the backward search of Graphplan, a disjunctive planner; and (3) evaluation of the utility of different types of mutex propagation routines in the context of SAT and CSP encodings of the automated planning problems. The results of this research have been disseminated to the planning community through three publications in the 5th International Conference on Artificial Intelligence Planning Scheduling and one publication in the 2000 National Conference on Artificial Intelligence.
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 01, 2001
- Accession Number
- ADA398018
Entities
People
- Sabbarao Kambhampati
Organizations
- Arizona State University