Résumé : Signaling is a key capability in collective behaviors for robot swarms. Yet, the current literature on the automatic design of robot swarms does not provide an approach that effectively leverages signaling capabilities in a generally applicable way. In this thesis, I propose that automatic modular design (AutoMoDe) is a suitable approach to addressing this gap. I aim to show that AutoMoDe can leverage environmental and inter-robot signaling to automatically generate the control software necessary for robot swarms to operate effectively. I investigate this issue in the context of design problems where a robots must cooperate, react to events, and perform missions sequentially. The presented research shows that AutoMoDe can effectively leverage signaling to address a wide range of design problems, including designing spatially-organizing behaviors, stigmergy-based behaviors, shepherding behaviors, and the design of robot swarms by demonstration.