SOCIAL ENGAGEMENT WITH ROBOTS AND AGENTS: INTRODUCTION

Engaging people in interactions with robots and agents is easy: it has been shown repeatedly that users respond socially to them. Keeping people engaged over a longer period of time is much more difficult—but this is exactly what companions or socially assistive robots should be able to do.

The concept “companion” has become an umbrella concept for a loosely defined collection of virtual (agent) or embodied (robotic) artifacts mainly for household and personal use (cf. Wilks 2010). They can be tutors, coaches and assistants in different domains, entertainers, and simply “company.” Experiments so far have considered scenarios such as weight loss and fitness training (Kidd 2008; Bickmore et al. 2005; Bickmore and Picard 2005), and pet substitute (like Paro). In the framework of “Ambient Assisted Living,” companions could play a role in organized home management, facilitation of social networking, and health monitoring, through to the physical tasks executed by mobile “social assistive robots” (SAR) (Feil-Seifer and Mataric 2005). Ideally, a companion should serve most or all of these functions in one application as a multitalented “virtual butler” (cf. Trappl 2011).

The idea is that people keep their companion with them for years, that it blends seamlessly into their everyday life, adapts to its user by learning and even manages to follow the user's changes in lifestyle, well-being, activities, and contacts. First of all, interaction with it has to be rich and variable enough to keep the user interested over a long period of time. But this is only the first of several challenges.

The second challenge is the intricacy of the social setting. An assistive companion would have different tasks and services. These services are not all of the same nature, such as the different stations to choose from on a radio. The companion does not only sit (or stand) there, waiting to be called into service by the user; it has its “own” agenda to pursue. Its very own needs would be those that have to do strictly with its functionality and survival, but by its own needs in a broader sense we understand those tasks that it executes on behalf of institutions that it—also or in the first place—serves. Entertainment companions may be completely their users' own, but the assistive companion functions in a social setting which includes the user and one or more institutions (e.g., medical and healthcare, caregivers, family, senior citizen association). Aaron Sloman (2010) makes the point succinctly in warning against calling the user the companion's “owner” lightheartedly, without examining what the real ownership relations are. The companion may be installed “by prescription” to run an application that monitors physical rehabilitation. Simultaneously, it could have a service that reminds the users to measure and report their blood pressure, or to take their medication. In yet another application of the same companion, family of the independently living older person request and get information about the well-being of the user and are alerted whenever something is amiss. And finally, the senior citizens' club uses its members' companions to announce events and to motivate users to participate in social activities. Add to these the tasks of information, reminders, and entertainment on the user's agenda, and the companion will face a considerable coordination and evaluation problem. The core challenge, however, is that not all of the companion's tasks are welcome or opportune at all times. Reminding, persuading, and giving advice are highly socially sensitive activities, as we all know from experience. It happens that we get angry at being reminded, although we know quite well that the person who is reminding us is right. Matters become even more complicated when we approach the companion's domain of autonomous action: for example, to call in caregivers, family, or medical support when the user is not compliant with the health regime, or when an emergency situation is detected. Aside from the obvious ethical questions touching on privacy, autonomy, and dignity of users, the challenge for research is to build and maintain human-companion relationships that allow both parties to achieve their goals on friendly terms.

A third challenge to mention is the environment in which the companion must function. It is not the fabricated world of the lab nor “robot house” (e.g., at the University of Hertfordshire, http://adapsys.feis.herts.ac.uk/), from which not only the objects not pertaining to the experiment have been removed, but also the persons with their everyday activities and social practices. The world in which a companion has to function is the private home that is created and run by the inhabitants' needs and practices.

We do not know enough about these challenges to design good assistive companions yet. The project SERA (Social Engagement with Robots and Agents, EU FP7, contract no. 231868) made one step further toward gaining insights about life with companions. The present special issue brings together results of and reflections on the research work undertaken in the project during the years 2009 and 2010. The project's core activity was a field study with a robotic interface (the Nabaztag) in health-related contexts. The description of the setup and the field study is provided by the article, Describing the Interactive Domestic Robot Setup for the SERA Project.

The article Theory of Companions: What Can Theoretical Models Contribute to Applications and Understanding of Human-Robot Interaction? begins, as did the project, with a review of social psychological theories of human-human relationships. Motivated by the data from the field study, the authors then critically ask to what degree and in which respects human-human relationships could and should serve as models for social robots, or whether we will rather need specific theories of human-robot interaction based on empirical research.

In addition to the field study “in the wild,” studies under controlled conditions were done with the same robotic-user interface. The results are presented in the article, Empirical Results on Determinants of Acceptance and Emotion Attribution in Confrontation with a Robot Rabbit.

We were able to collect over 300 video recordings of consenting participants, showing them in interaction with their robotic companion in their homes, over a period of approximately ten days. Observational data of this kind are a rare and nearly undprecedented resource for research. Different project teams from different disciplines worked on the analysis of this material (HCI, psychology, engineering, ethnomethodology) with their respective “tools of the trade.” Reflections on methodology in analyzing human-robot interaction thus were an integral part of the project work. The article, From Data to Design, addresses this issue from the engineers' perspective. It takes a critical look at qualitative methods of data analysis and proposes the narrative method as a path from episodic data to design.

The final article, On the Nature of Engineering Social Artificial Companions, carries on the development perspective. It first discusses the nature of requirements engineering in the field of assistive companions. In its second part, it analyzes in depth a problematic (and recurring) interaction sequence from the data and presents a model for the turn-taking and topic-management skills that are necessary to deal with this type of interaction.

The conclusion reached by the last article is valid for the all of the articles: in order to build artificial social companions, we need a thorough understanding of the assumptions that humans and machines each make about the assumptions of the other. One aspect of this is an understanding of what stance the human takes toward the technology, and the other is what the technology assumes about the human in interaction. To arrive at an understanding of how artificial companions work, it is therefore necessary to study real interactions between humans and machines in ecologically valid situations. The statement closes the circle of this collection of articles, referring back to the article about the field study, which now can be seen as an effort to implement such a study, not just to be repeated, but to be extended and improved through further research.

For more details about the project and its results—which include an interactive showcase and tools for interaction design—the reader is invited to visit the website http://showcase.project-sera.eu/.

Acknowledgments

The research leading to these results has received funding from the European Community's Seventh Framework Programme [FP7/2007-2013] under grant agreement no. 231868 and from the Austrian Federal Ministry of Science and Research under its Additional Funding Programme.