Representing Knowledge within the Situation Calculus Using Interval-Valued Epistemic Fluents

Abstract

The ability of interval arithmetic to provide a finite (and succinct) way to represent uncertainty about a large, possibly uncountable, set of alternatives turns out to be useful in building "intelligent" autonomous agents. In particular, consider the two important issues of reasoning and sensing in intelligent control for autonomous agents. Developing a principled way to combine the two raises complicated issues in knowledge representation. In this paper we describe a solution to the problem. The idea is to incorporate interval arithmetic into the situation calculus. The situation calculus is a well known formalism for describing changing worlds using sorted first-order logic. It can also be used to describe how an agent's knowledge of its world changes. Potentially, this provides a sound basis for incorporating sensing into logic programming. Previous work has relied on a possible worlds approach to knowledge. This leads to an elegant mathematical specification language. Unfortunately, there have been no proposals on how to implement the approach. This is because the number of possible worlds is potentially uncountable. We propose an alternative formalization of knowledge within the situation calculus. Our approach is based on intervals. The advantage is that it is straightforward to implement. Moreover, we can prove that it is sound and (sometimes) complete with respect to the previous possible worlds approach.