Qualitative Spatial Reasoning

Rules play an important role in everyday human interaction. In road traffic the participants have to obey to the rules to guarantee smooth traffic flow and to avoid accidents. The participants in these situations are called agents. Artificial agents that act in this environment must be aware of these rules and must act accordingly. Typically, such rules are formulated in natural language and thus, contain qualitative terms like “from left” or “turn right”. These terms are only intricately to implement to navigate agents. In this article we describe a so-called qualitative approach to formalize natural language spatial regulations. With the demonstrator SailAway [1] we show how the formalizations can be applied with little computational effort for collision free navigation in the domain of vessel navigation.

Autonomous vehicles are employed in intralogistics in order to transport work pieces between different workplaces. Previous guidance systems are based on wire-guided tracks or optical following of surface markings. In the first case, the possibility to change the production layout is rather limited. In the second case, abrasion can significantly decrease recognition rates. In contrast, humans easily succeed in navigation tasks, even in dynamic environments. A promising approach is therefore to apply a cognitively motivated spatial representation for autonomous vehicles. This article presents BA23, a set of 23 qualitative relations for qualitative spatial reasoning, and discusses its application in intralogistics.