Sea waves are a source of energy with a considerable estimated worldwide potential, but no large-scale installations are operating yet. One of the issues of wave energy when compared with other renewable sources is the limited scalability of device sizes. Since devices are relatively small, installation in parks is a mandatory choice. Closely spaced devices interact hydrodynamically with each other, possibly unfavourably in terms of energy production. Hence, determining the positions of devices so that production is optimized is an important design step. Moreover, the sea environment has large variability, thus subjecting devices to a large set of conditions, including extreme events. This requires a robust approach to design and control. The topic of this talk will be our work on robust optimization of parks of wave energy converters. First, we focused on robust optimization of control variables while considering the positions fixed. We worked in the frequency domain, under the hypotheses of the linear potential flow model (Airy waves) and solving the hydrodynamic problem using the boundary element method. To model uncertainty, the direction of ambient waves was treated as a random variable, while the frequency dependence of the spectrum was dealt with deterministically. Two different approaches to robust optimization were considered: stochastic gradient descent and the sample average approximation. A comparison between the two strategies will be presented, together with results of engineering interest. This is joint work with G. Ciaramella, E. Miglio and T. Vanzan.