The explicit representation of mental states, such as belief, desire, intention, etc., is a crucial issue for the design of multi-agent systems aimed at performing complex tasks which require some form of global intelligent behavior. In particular, methods for conflict resolution between competing mental processes are a key factor in determining the dynamic behavior of an autonomous agent. In this paper, a novel approach to model agent mental activity based on the concept of active mental entity is proposed. Then, the general organization and operation of a multi-agent architecture encompassing an explicit representation of agent mental activity is introduced. In this context, the issue of conflict resolution at the level of active mental entities is faced and suitable conflict resolution methods are proposed. Finally, an application example concerning a mail delivery robot is presented and discussed.
PDF file