This paper presents a technique for using a priori contextual knowledge for performing situation assessment for autonomous underwater vehicles (AUVs). What sets this technique apart from other assessment techniques is that it uses explicitly represented contextual schemas to describe discrete contexts that may occur in the world. A modified version of the internist-1/caduceus [10] algorithm is then used to diagnose the situation as an instance of one or more of the set of known contexts. The schemas representing these contexts can then be merged to give a coherent assessment of the current situation that can serve as the basis for the AUV's behavior.
展开▼