An Agent Program Logic with Declarative Goals

It has been argued that declarative goals provide for a natural conceptual tool for designing as well as programming agents. This has given rise to various proposals for integrating declarative goals into programming languages. It is not always clear, however, how to estahblish a precise relation to logical agent theories to reason about such agents. In this paper, we propose an agent programming theory that provides both an agent programming language as well as a corresponding agent program logic to verify agent programs. The agent programming language and agent progam logic are developed in parallel to ensure the existence of a mathematically precise relation between the program and logical semantics. To this end, a modal agent logic including the core agent concepts of action, knowledge and goals is introduced. Consecutively, an equivalent state-based semantics is introduced that can be used to relate the logical and operational program semantics. Additionally, it is shown how to integrate goal adoption into the programming theory using the concept of second-order goals.