A Case Study for Integrating Heterogeneous Knowledge Bases for Outdoor Environments

Abstract

We present the integration of heterogeneous knowledge bases and reasoning mechanisms in the SHERPA project. SHERPA is about a complex search and rescue scenario that requires planning and reasoning about actions of mixed human-robot teams in space and in time. This has been achieved by integrating different sources of knowledge and data, namely, OpenStreetMap for static map data, the Robot Scene Graph for composing the OpenStreetMap data with dynamic data from the environment and the rescue team, and KnowRob with the SHERPA ontology for abstract reasoning in the application domain. This work explains how these knowledge bases were integrated by model composition and model to model transformations and an application dependent bridge. Design decisions are discussed and an example is given that explains the usage of the heterogeneous knowledge and reasoning methods.