A Mereological Approach to Representing Spatial Vagueness

Spatialreasoningiscrucialin many AI applicationdomains, suchasrobotics,qualitativeand naive physics,and some types of planning. Qual- itativereasoningisoftenrequired; and in many of thesecases,uncertaintyorimprecisionabout thespa- tialextentofparticularentitieshas tobe represented and coped with.This paper developsan axiomatisa- tion ofa relationofcrisping'(reducingimprecision or vagueness)between pairsofvague'spatialregions: thosewith indeterminateboundaries.This axiomati- sationisthenrelatedtothepreviouslydeveloped'egg- yolk'representationofvagueness,originallydesigned by (Lehmann and Cohn 1994) fordatabase applica- tions,then appliedtoexpressingspatialvaguenessby (Cohn and Gotts 1994).