Dipartimento di Ingegneria informatica, automatica e gestionale

Aiming at improving our physical strength and expanding our knowledge, tournaments and competitions have always contributed to our personal growth. Robotics and AI are no exception, and since beginning, competitions have been exploited to improve our understanding of such research areas (e.g. Chess, VideoGames, DARPA). In fact, the research community has launched (and it is involved) in several robotics competitions that provide a two-fold benefit of (i) promoting novel approaches and (ii) valuate proposed solutions systematically and quantitatively. In this paper, we focus on a particular research area of Robotics and AI: we analyze multi-robot systems deployed in a cooperative-adversarial environment being tasked to collaborate to achieve a common goal, while competing against an opposing team. To this end, RoboCup provide the best benchmarking environment by implementing such a challenging problem in the game of soccer. Sports, in fact, represent extremely complex challenge that require a team of robots to show dexterous and fluid movements and to feature high-level cognitive capabilities. Here, we analyse methodologies and approaches to address the problem of coordination and cooperation and we discuss state-of-the-art solutions that achieve effective decision-making processes for multi-robot adversarial scenarios.