BDI : An Alternating-time BDI Logic for Multiagent Systems

Many logics for modelling beliefs, desires and intentions of agents, such as Rao and Georgeff’s BDI logic and Wooldridge’s LORA, are based on temporal logics like CTL/CTL∗ (Computational Tree Logic) in which the structure of time is branching in the future and linear in the past. Recently, many attempts have been made to define logics for BDI agents by using extensions of CTL. In this paper, we discuss BDI that is obtained by substituting ATL∗ (Alternating-Time Temporal Logic) to CTL∗ in Rao and Georgeff’s logic. One of the main advantages of our approach is that in BDI we can express new commitment strategies that are more realistic than those proposed by Rao and Georgeff (and that could not be defined in their logic), since they take collaboration among agents into account. In particular, in this paper we discuss three variants of Rao and Georgeff’s “open minded” commitment: “independent open minded”, “optimistic open minded”, and “pessimistic open minded”, whose definition exploits the new features that ATL∗ adds to CTL∗.