Dealing with variability may be the most serious issue when developing and maintaining robotic systems: it jeopardizes communication between the stakeholders of a robotic system and interoperability between its components. This paper reports on an approach and a toolset to improve both communication and interoperability. The approach relies on an ontology of mobile robotics and on a Domain-Specific Language (DSL) to describe robotic systems with missing parts that are problems that can be completed by third-party solutions. The domain model of the DSL is derived from the conceptual framework of the ontology. Models of robotic systems and scenarios can be defined using a graphical modeling environment and code can be generated for open-source as well as commercial middleware and simulators. A key advantage of the approach is that code can be generated for both simulation platforms and real-world robots. The validation of the approach is illustrated with a landmark-search case study.