One main topic of SG2 is the analysis and comparison of approaches to logics of action. This includes extensions of dynamic logic, modal-logic approaches, logics for causality and initiative, etc and maybe other approaches which will be discovered during the work in SG2. The contribution is mainly in the development of applied (or applicable) logics of action. This includes axiomatization, declarative semantics and operational semantics. Applicable logics of action should support both admissible and normative behavior since we have to deal with hybrid systems consisting of human and machine agents.
A second aspect is the relation of these logic approaches to models of actions in the context of agents. The notion of agent state connected to actions, models of communication and cooperation, aspects of distribution (``true'' concurrency with truly independent actions), visibility of and knowledge about other agents, behaviour evolution depending on communication history, etc are some aspects important for this topic.
Putting these aspects together, the main objective of SG2 should be collecting the various diverse logics and models of (concurrent) actions, and assessing/integrating them in the intended context(s) in order to arrive at a flexible and useful model of a multi-agent system.