This paper describes an approach for reasoning about the interactions of multiple agents in moderately complex environments.
The semantics of Belief Desire Intention (BDI) agents has been investigated by many researchers and the gap between theoretical specification and practical design is starting to be bridged. However, the research has concentrated on single-agent semantics rather than multi-agent semantics and has not emphasised the semantics of the environment and its interaction with the agent.
This paper describes a class of simple BDI agents and uses a recently introduced logic of actions to provide semantics for these agents independent of the environment in which they may be embedded. The same logic is used to describe the semantics of the environment itself and the interactions between the agent and the environment. As there is no restriction on the number of agents the environment may interact with, the approach can be used to address the semantics of multiple interacting agents.