PhD defence Michiel Joosse

investigating positioning and gaze behaviors of social robots 

As technology advances, application areas for robots are no longer limited to the factories where they perform repetitive tasks behind fences. Robots are envisioned to provide services to us in everyday public spaces - in which they will encounter and interact with people. These types of robots can be considered as social robots, and should interact with people following the behavioral norms people expect of the robot within a specific context.

Compared to interactions people have with social robots in homes and classrooms, interactions with guide robots in public spaces are likely to be of a more incidental nature and shorter duration. This makes it perhaps even more important that users immediately understand how to use the robot. And what intentions and messages the robot communicates. Given that people apply social rules when interpreting the behavior of media (such as computers and robots), we have investigated social norms for social robots; in particular the non-verbal behaviors of (interpersonal) distance and gaze.

In this thesis we have investigated people's preferences for, perceptions of, and behaviors towards social robots through a series (lab and field) studies.

We first conducted a contextual analysis to better understand the future operating environment of the robot. Together with a literature overview of related work in the field of Human-Robot Interaction, the insights we gained the contextual analysis guided the remaining work in this thesis. To address questions about a group's walking speed while carrying luggage we conducted the second study. The results of the first two studies have been used in the design of the third study, in which we investigated people's expectations of a guide robot's behavior in conflicting social situations as it guides small groups.

The next three studies focused on people's perceptions of a robot's approach behavior when approaching a small group of people. The main findings of these three studies are that people adapt the spatial arrangement of their group to the robot. Furthermore, we investigated the perception of approach distances- and directions of Chinese, Argentinian and United States participants. Our results indicate that Chinese participants had a preference for closer approaches over farther approaches. Preferences from people from the United States were the opposite. Finally, we found that projective measures could be used to explore changes in human-robot personal space preferences and behaviors when manipulating an independent variable.

In the seventh and eighth study of this thesis, we investigated people's understanding of a robot's head direction- and turn behavior while the robot guides a small group of people.  We found that a robot's head should face the driving direction when guiding a small group of people, and that occasionally turning towards the people following the robot is interpreted as ``caring for them". Finally we conducted a case study in which we investigated under which circumstances the current state-of-the-art in robotic technology could provide added value to airport passengers. This study shows that a social robot at an airport would be a useful service for in particular for people inexperienced with air travel, or to provide information in situations where disruptions of normal operations occur.

The work presented in this thesis provides empirical and methodological contributions to the development of social robots in semi-public spaces. The research in this thesis shows that the behaviors of social robots that provide services such as wayfinding in public spaces (such as airports) should, to an extent, be designed in a human-like way. This thesis offers practical guidelines for designers and developers of social robots. Specifically in terms of which distance robots should keep from groups of people, and in which direction a robot should gaze when guiding (small) groups. The results of this thesis contribute toward the development of social robots in semi-public spaces. The behavior of social robots should suit the context of use. In order to design behaviors for future social robots it is necessary for HRI researchers to study people's interactions with robots in this context of use during the development cycle of a social robot.