Answer Set Programming to Model Plan Agent Scenarios

Abstract

One of the most challenging aspects of reasoning, planning, and acting in an agent domain is reasoning about what an agent knows about their environment to consider when planning and acting. There are various proposals that have addressed this problem using modal, epistemic and other logics. In this paper we explore how to take advantage of the properties of Answer Set Programming for this purpose. The Answer Set Programming's property of non-monotonicity allow us to express causality in an elegant fashion. We begin our discussion by showing how Answer Set Programming can be used to model the frog’s problem. We then illustrate how this problem can be represented and solved using these concepts. In addition, our proposal allows us to solve the generalization of this problem, that is, for any number of frogs.