The development of software applications for robotics is a challenging task for engineers due to the inherent parallelism of the code and the multitude of flexible factors that must be considered. Robotic assembly, a specific type of robotic application, involves the construction of pre-determined products using designated pieces. This further complicates the engineering task, as not only the robots must be considered but also the construction materials and the product itself. While several specialized solutions for robotic assembly exist, they are typically handcrafted and are tightly bound to the robots and materials they were designed for. Thus, it is hard for these solutions to be extended to respond to changes and newly added modules. This work proposes a model-driven approach to robotic assembly supporting diverse domains that can accommodate the seamless integration of robots. Relational reference attribute grammars are used for both metamodels and runtime models. To assess the feasibility and effectiveness of our proposed approach, we conducted a case study based on an industrial research project.