Rule-based reasoning about qualitative spatiotemporal relations

Abstract

This paper is about a novel rule-based approach for reasoning about qualitative spatiotemporal relations among technology-rich autonomous objects, to which we refer to as artifacts. The objective of our work is to provide means for defining spatiotemporal constraints - i.e. logical combinations of spatial relations to artifacts at certain time intervals - at a high level of abstraction, and to recognize relative situations therewith. Such constraints are defined with rules that infer high-level relationships for newly recognized situations, which in turn can be used in other constraints. At any time, the history of known relationships can be queried in order to trigger predefined actions. We decided for qualitative abstractions of both spatial and temporal relationships, as they reflect the semantics of natural language terms and thus facilitate dealing with relationships at the application programming level. The core concepts of this reasoning approach are presented, and the implementation of a middleware for spatiotemporal reasoning as well as evaluation results are discussed.