Human-robot interaction (HRI) is an important area of robotics research; it is related to human-computer interaction (HCI), but contains a key difference: HRI allows embodied systems to utilize physical context and mobility. Most current HRI systems, however, do not yet utilize mobility for interactive purposes. In this paper, we describe the design and evaluation of a robot system aimed at embodied HRI communication. The goal of the system is for the robot to select a target for interaction based on the perceived desire of human subjects to interact with it. We use an interaction policy inspired by Joint Intention Theory to shape the interaction between the users and the robot. The eventual purpose of the robot system is for use in crowded classrooms for selecting students most desirous of interaction and help.