The maritime industry is undergoing a rapid transformation driven by decarbonisation and digitalisation, leading to increasingly integrated and complex onboard systems. Ensuring the robustness of these systems, defined here as a ship’s ability to maintain vital functions despite random component or connection failures, is critical, particularly during early design stages where modifications are less costly. This paper presents a novel robustness evaluation method adapted to ship concept design, utilising network theory to model interdependencies across electrical, control, and data acquisition systems. The method incorporates metrics adapted to maritime engineering, enabling efficient assessment of multiple system configurations and identification of vulnerable components. Applied to five inland vessel concepts varying in automation levels, the results confirm that redundant control options substantially improve robustness, and that elasticity-based metrics provide consistent robustness comparisons across networks of differing sizes. Conversely, classical metrics such as natural connectivity and effective graph resistance potentially exhibit sensitivity to network size, limiting their applicability. The study emphasises the importance of integrating robustness evaluation into early design processes to inform the development of safer, more reliable autonomous vessels. It highlights potential directions for future research, including enhanced failure modelling and statistically based simulations.