Reactive policies with planning for action languages

Abstract

Action languages are an important family of formalisms to represent action domains in a declarative manner and to reason about them. For this reason, the behavior of an agent in an environment may be governed by policies which take such action domain descriptions into account. In this paper, we describe a formal semantics for describing policies that express a reactive behavior for an agent, and connect our framework with the representation power of action languages. In this framework, we mitigate the large state spaces by employing the notion of indistinguishability, and combine components that are efficient for describing reactivity such as target establishment and (online) planning. Our representation allows one to analyze the flow of executing the given reactive policy, and lays foundations for verifying properties of policies. Additionally, the flexibility of the representation opens a range of possibilities for designing behaviors.