Modeling Abstract Behavior: A Dynamic Logic Approach

Modeling abstract behavior is essential for intelligent agents under incomplete and uncertain environments. In this paper, we extend Propositional Dynamic Logic (PDL) to Propositional Abstract Dynamic Logic (PADL) for modeling abstract behavior in two aspects. On the one hand, we treat the task of finding a plan to achieve a certain formula as an abstract action. On the other hand, we explicitly represent the subsumption relation between two actions as a formula in the language. We propose the semantics for the two operators and discuss some important related properties.