Modal logics for AI planning

This paper presents a survey on current research in modal and dynamic logic in the theory of AI planning techniques together with a first assessment of the role these concepts and techniques may play in the overall process of automated plan formation. First, the use of modal logics for modelling planning problems is motivated. After showing the computational problems-besides the well known frame problem-of the situation calculus by an example, it is argued that a modal framework for modelling planning problems offers several advantages compared with classical planning approaches. Then, the basic ideas of modal logics are presented, and two existing approaches to planning using a modal framework are reviewed. At the end, an outlook towards efficient implementations of modal logics is given.<>