To this end, this article proposes a novel approach named sequential MARL with role assignment using transformer (SMART). By learning the effects of different actions on state transitions and rewards, ...