Charles Explorer logo
🇬🇧

Constraint Based Reasoning over Mutex Relations in Graphplan Algorithm

Publication at Faculty of Mathematics and Physics |
2006

Abstract

This paper describes a work in progress. We address a problem of reachability of states in planning from the constraint programming perspective.

We propose an alternative way how planning graphs known from the Graph-plan algorithm can be constructed. We propose to do an additional consistency check every time when our variant of the planning graph is extended.

This leads into a more accurate approximation of reachable states than it is done by the standard Graphplan algorithm. It is theoretically shown in this paper that our variant of planning graph forbids strictly more situations than it is forbidden by the standard planning graphs.