Robots are finding their ways into our everyday environment (68) and as such their social aspect and interaction with humans are becoming a very important issue for their successful deployment and control. However, "directly" interacting with robots has thus far been largely unsuccessful. One main reason is that most robots are not (or perhaps cannot be) designed well human factor-wise for diverse users. Figure 6.1 illustrates the case very well. Touch screens (on the robot) are usually small and difficult to reach, and their interfaces are difficult to use. A robot as a moving desktop computer may not be a well thought out metaphor despite our familiarity with the desktop interfaces. On the other hand, today's robots tend to pursue human-like communication means while still lacking much in most such abilities. Robust vision-based or voice/language-based robot communication seems very difficult in practice. The problem of anthropomorphic interfaces (of which robots are usually) generating false expectations of human-like responses and behaviors from robots, and becoming a source of disappointment and confusion to users, has been well known (18; 122). Figure 6.2 shows shots from the movie Star Wars (3) in which the robot R2D2 uses a 3D holographic projection to convey a message from Prince Leia. Even though the movie is only an imaginary plot, it is quite plausible to suggest that such "non-human-like" communication means can be more effective, at least for certain tasks or situations. Interestingly, in this particular shot, C3PO, a perfectly human-like robot, also listens in to the projected message. While robotics research must be continued in the direction to produce human-like capabilities, we propose (inspired by ideas from StarWars) an alternative, that is, an "indirect" human-robot interaction (HRI) mediated through a hand-held device and a projection display (since 3D hologram technology is still immature). Figure 6.3 shows one possible way to employ such a framework where the robot projects onto a donut shaped area on the floor to serve and interact with multiple users at the same time. Such an HRI method could be useful for large but sparsely populated areas like conventions, airports, post offices, etc. The robot can move around and stop to offer interactive services by using a camera (and/or other sensors) to detect anybody in the vicinity who may be interested to interact. Figure 6.4 shows another possibility in which a robot actively guides a group of people in a museum. The robot, given the map of the museum and a predetermined path, with access to localization information for itself and the people surrounding it (using onboard sensors or infrastructure), can select an appropriate surface to project information for the users We claim that such an interaction framework can alleviate much of the human factors problem and enhance the overall experience by the "enlarged" communication channel and amplified presence of the robot. In addition, the proposed framework is flexible to accommodate different types of users and tasks, one-to-many interaction, even mixing of real and virtual objects, and when combined with a hand-held interaction device, it can be extended to support a variety of interactiontechniques with the consideration of robot's mobility. In this paper we concretize our proposal and discuss various important technical issues in realizing it.
ASJC Scopus subject areas
- General Computer Science