Systems with knowledge representation and reasoning functionality are quite common within the robotics community. Nevertheless, most of the proposed architectures are ad-hoc implementations, which lack of modularity, standardization, and that cannot be reused or shared among different users. We fill that the Semantic Web effort over the past years in promoting standard representations and ontology best practices should be adopted by the robotic community, to foster the design of more effective and reusable systems, adapt for non expert users in everyday activities. In this paper, we investigate how to integrate ontology best practices and standard representations into the robotic system design, modelling an ontology for a real application field: urban search and rescue robotics. We also discuss the benefits of this methodology for robotic systems, as well as presenting some effective design guidelines.