Detecting Physical Defects: A Practical 2D-Study of Cracks and Holes

No. 2: Benchmarking of Qualitative Spatial and Temporal Reasoning Systems

Volume

Papers from the 2009 AAAI Spring Symposium

Contents

Theoretical work on Qualitative Spatial Reasoning (QSR) is abundant, but the actual requirements of practical applications have been widely ignored. This paper discusses how ontologies allow to compare different QSR formalisms with respect to definability of spatial concepts, which are taken from a real-world problem. We introduce the problem of detecting phys- ically defective parts (such as in manufacturing) and review which qualities are necessary for modeling these as QSR problem. We show that —besides standard mereotopological con- cepts — a set of artifacts, especially cracks and holes, are of foremost importance in the domain of interest. However, most currently available region-based QSR approaches fail to distinguish these. In the future, the proposed set of problem can be used to evaluate different QSR formalisms for their adequacy with respect to defining and distinguishing cracks and holes.