Opportunities: a unifying framework for planning and execution

摘要

A successful agent in the real world must both plan ahead and react to the unexpected. Ideally, both processes should be carried out in a common framework. In this paper we describe such a framework based on the analysis of opportunities. We argue that planning in advance can be viewed as a matter of anticipating opportunities, while responding to the unexpected should be seen as reacting to opportunities when they arise. We present an opportunistic planning agent, PARETO, that operates in a simulated robot delivery world, and implements our approach.