Dynamic Flexible Constraint Satisfaction and it's Application to AI Planning

Abstract

Constraints are a natural means of knowledage representation in many disparate fields. A constraint often takes the form of an equation or inequality, but in the most abstract senseis simply a logical relation among several variables expressing a set of admissable value combinations. The following are simple examples: the sum of two variables must equal 30; no two adjacent countries on the map may be coloured the same. It is this generality and simplicity of structure which underly the ubiquity of the constraint-based representation in Atificial Intelligence.