From the Specification of Multiagent Systems by Statecharts to their Formal Analysis by Model Checking

A formalism for the specification of multiagent systems should be expressive and illustrative enough to model not only the behavior of one single agent, but also the collaboration among several agents and the influences caused by external events from the environment. For this, state machines [25] seem to provide an adequate means. Furthermore, it should be easily possible to obtain an implementation for each agent automatically from this specification. Last but not least, it is desirable to be able to check whether the multiagent system satisfies some interesting properties. Therefore, the formalism should also allow for the verification or formal analysis of multiagent systems, e.g. by model checking [6]. In this paper, a framework is introduced, which allows us to express declarative aspects of multiagent systems by means of (classical) propositional logic and procedural aspects of these systems by means of state machines (statecharts). Nowadays statecharts are a well accepted means to specify dynamic behavior of software systems. They are a part of the Unified Modeling Language (UML). We describe in a rigorously formal manner, how the specification of spatial knowledge and robot interaction and its verification by model checking can be done, integrating different methods from the field of artificial intelligence such as qualitative (spatial) reasoning and the situation calculus. As example application domain, we will consider robotic soccer, see also [24, 31], which present predecessor work towards a formal logic-based approach for agents engineering.