Planning and Plan Recognition

Abstract

This paper introduces the reader to the areas of automatic planning and plan recognition. The former concentrates on the goal-directed synthesis of plans from primitive actions, and has applications in robotics, experiment design, and other fields. The latter concentrates on methods for inferring the goals that underlie an agent's actions, in order to support more natural human/machine interfaces and to model human communication. The paper also provides an overview of some recent results obtained by the authors as part of their doctoral research. The authors are continuing research in these areas at AT&T Bell Laboratories.