Maintenance activities on transportation infrastructures, particularly on the underside of bridges and viaducts, present critical challenges related to safety, accessibility and cost-efficiency. Conventional solutions, such as scaffolding or mobile platforms, often exhibit substantial operational limitations. In this context, modular articulated robotic arms present a promising, safer, and more flexible alternative. This study introduces a functionality and data-driven design methodology for the modular development of articulated robotic arms specifically tailored for bridge maintenance applications. The proposed approach integrates Finite Element Analysis with clustering algorithms to identify mechanically equivalent joint modules based on their structural response under realistic loading conditions, including self-weight and external forces. Subsequently, topology optimisation techniques are applied to refine the geometry of selected joints, enhancing their stiffness-to-weight ratio while preserving manufacturability and modularity. Additionally, the application of Geometric Dimensioning and Tolerancing ensures mechanical reliability and precision during assembly. The effectiveness of the methodology is demonstrated through comparative Finite Element Analysis assessments between the initial and optimised designs, along with the construction of a functional prototype. The proposed framework paves the way for scalable, robust, and repeatable robotic solutions in the infrastructure maintenance sector.