Autonomous underwater robots, such as the UX-1 developed in the UNEXMIN project, need to maintain reliable autonomous operation in hazardous and unknown environments. Because of the lack of any kind of real-time communications with a human operated command and control station, the control architecture needs to be enhanced with mission-level self-diagnosis and self-adaptation properties an additional provided by some kind of supervisory or “metacontrol” component to ensure its reliability. In this paper, we propose an ontological implementation of such component based on Web Ontology Language (OWL) and the Semantic Web Rule Language (SWRL). The solution is based on an ontology of the functional architecture of autonomous robots, which allows inferring the effects of the performance of its constituents components in the functions required during the robot mission, and generate the reconfigurations needed to maintain operation reliably. The concept solution has been validated using a hypothetical set of scenarios implemented in an OWL ontology and an OWLAPI-based reasoner, which we aim at validating by integrating the metacontrol reasoning with a realistic simulation of the underwater robot.