In the last three years, several ``radically new'' and promising approaches have been developed for tackling the plan synthesis problem. Currently, these approaches exist in isolation as there is no coherent explanation of their sources of strength vis a vis the traditional refinement planners. In this paper, I provide a generalized view of refinement planning, that subsumes both traditional and newer approaches to plan synthesis. I will interpret the contributions of the new approaches in terms of a new subclass of refinement planners called disjunctive planners. This unifying view raises several intriguing possibilities for complementing the strengths of the various approaches. I will identify and pose these as challenges to the planning community.
An html version of the paper.
Here is the link to the
postscript version of the full paper (it is now portable)
Request for comments and participation
Page accessed
times
Subbarao Kambhampati
Last modified: Sat Nov 20 09:36:11 MST 2004