The aim of our study is to investigate systems for supporting remote instruction via a mobile robot. In the real world, instructions are typically given through words and body orientations such as head movements, which make it possible to project others' actions. Projectability is an important resource in organizing multiple actions among multiple participants in co-ordination with one another. It can likewise be said that in the case of robot-human collaboration, it is necessary to design a robot's head so that a local participant can project the robot's (and remote person's) actions. GestureMan is a robot that is designed to support such projectability properties. It is argued that a remote controlled mobile robot, designed as a communication medium, makes relevant dual ecologies: ecology at a remote (robot operator's) site and at a local participant's (robot's) site. In order to design a robot as a viable communication medium, it is essential to consider how these ecologies can be mediated and supported.

In this paper we investigated systems for supporting remote collaboration using mobile robots as communication media. It is argued that the use of a remote-controlled robot as a device to support communication involves two distinct ecologies: an ecology at the remote (instructor's) site and an ecology at the operator's (robot) site. In designing a robot as a viable communication medium, it is essential to consider how these ecologies can be mediated and supported. In this paper, we propose design guidelines to overcome the problems inherent in dual ecologies, and describe the development of a robot named GestureMan-3 based on these guidelines. Our experiments with GestureMan-3 showed that the system supports sequential aspects of the organization of communication.