Current students

Recently, the study of environments in which multiple agents operate has been gaining increasing interest among the Artificial Intelligence community. This growth of the field of Multi-Agent Learning (MAL) led to important scientific breakthroughs such as the development of superhuman bots for the games of no-limit Texas hold'em poker (Libratus and Pluribus) and StarCraft II (AlphaStar). However, some fundamental challenges of multi-agent environments still remain not tackled. This project aims at studying some of such challenges that emerge when considering the interaction between software agents and humans. In particular, we focus our interest in those situations in which the strategy space suffers from a combinatorial growth in its size and in which the objective of the agents are not only the maximization of their expected reward but also the satisfaction of some safety constraints.