Query-based integration of heterogeneous knowledge bases for search and rescue tasks

Abstract

Recently, advances in robotics’ technology and research focus on complex scenarios. In these scenarios, robots have to act and respond fast to situational demands. First, they require heterogeneous knowledge from various sources. Then, they need to integrate this knowledge with their reasoning methodologies. These reasoning methodologies are typically different for every domain. This paper introduces an integrated knowledge processing methodology. This methodology uses query mechanisms and model-to-model transformations. Combining these two mechanisms enables processing of heterogeneous knowledge bases. The methodology is demonstrated for an outdoor scenario with diverse systems. In this scenario knowledge and reasoning methods from various sources are integrated. This includes static knowledge from. Open Sreet Map and Digital Elevation Models. The Robot Scene Graph tracks changes in the world and provides geometric reasoning. KnowRob with its Sherpa ontology and openEASE provide further reasoning capabilities.