Skip to main content
European Commission logo print header

Embodied Cognitive Models for Fluent Human-Robot Interaction

Final Report Summary - HUMAN ROBOT FLUENCY (Embodied Cognitive Models for Fluent Human-Robot Interaction)

Robotics is rapidly moving from industrial applications into personal, office, health care, and shop environments, necessitating research in human-robot interaction. The goal of this project was to investigate the role of nonverbal and embodied aspects of human-robot collaboration and companionship. This is due to the fact that humans are extremely sensitive to nonverbal cues, and in particular to their timing.

The project resulted in the design and construction of several physical robotic companions and simulated systems for studying the nonverbal aspects of human-robot interaction.

Through this project, Dr. Hoffman deepened our understanding of the relationship between timing collaboration. In particular, he established - for the first time - the relationship between objective measures of team timing, and people’s subjective sense of a robotic team member. This work was awarded a Best Workshop Paper in 2013.

Understanding the minutia of human-robot timing is crucial if we are to integrate robots into our workforce. Building robots that display appropriate timing behavior, and being able to objective measure their success, will be key for the large-scale implementation of robotic collaborators in our workforce.

Beyond human-robot labor collaboration, Dr. Hoffman also contributed to the development of new paradigms for designing personal robotic companions, both for assistive applications and for the home.

He pioneered a design process, “Movement-centric design”, which has led to the development and physical construction of three new socially expressive robotic companions, Travis, Kip, and Vyo. The paper describing the design and evaluation of Kip was awarded a Best Paper award in the field’s most competitive conference, the IEEE/ACM International Conference for Human-Robot Interaction in 2015.

In addition, Dr. Hoffman introduced the use of smartphones as computational platforms for low-cost, cloud-connected social robots. The paper describing this approach was awarded a Best Paper nomination at the IEEE International Symposium on Robot and Human Interactive Communication in 2012.

Dr. Hoffman used these new robots in experimental research conducted in collaboration with the IDC School of Psychology and a number of top universities in the US, including Carnegie Mellon University, the Georgia Institute of Technology, Duke University, Northwestern University, and the University of Rochester.

In two studies conducted as part of this research, the research team found that a robot responding nonverbally to an event can make that event more acceptable to the human, and also has positive effects on the human's perception of the robot as an agent. This has applications for robots sharing our spaces, such as homes, offices, and care facilities.

Another study evaluated the situation of a robot listening to people telling them about a difficult event in their lives. The research team found that the nonverbal and textual responsiveness behavior affected people's evaluation of the robot, in terms of competence, appeal, and sociability. As responsiveness plays a major role in any interaction that involves effective communication and social support, this research has important outcomes for robots in a variety of caregiving roles, such as nursing, childcare, education, and elder care.

With his collaborator at IDC, Dr. Oren Zuckerman, Dr. Hoffman also developed the new notion of “Empathy Objects”, which are peripheral robotic companions accompanying human-human interaction. Peripheral robotics provides for a key new research direction, as it doesn’t just study the traditional interaction between a human and a robot but the effect of a robot next to human interaction. This will be of increasing importance as robots become more prevalent in human spaces.

In one study, the robot was designed to promote non-aggressive conversation between people. It monitored the conversation’s nonverbal aspects and maintains an emotional model of its reaction to the conversation. The study evaluated the effects of the robot’s autonomous behavior on couples’ conversations, using both objective and subjective metrics. We found that a conversation companion reacting to the conversation led to more gaze attention, but not more verbal distraction. Participants also rated the reacting robot as having significantly more social human character traits and as being significantly more similar to them.

Using an additional new robotic head which was designed and build in collaboration with Carnegie Mellon University, Dr. Hoffman also conducted research related to robots in the workplace, in particular with respect to the effects of robots on people’s ethical behavior. As part of this project, Dr. Hoffman and his collaborators conducted two human-robot interaction studies.

One laboratory study (n=60) examined the effect of the robot’s presence and nonverbal behavior on participants tendency to not comply with instructions when they stood to receive financial gain from not complying. We found that the robot monitoring people’s behavior through social gaze behavior could increase compliance as much as a human presence.

Dr. Hoffman’s team followed up with a field study evaluating robot-induced honest behavior in the real world. This study used the same robotic head, mounted on a mobile robot, and was conducted at Carnegie Mellon University. Initial results indicate the robot’s presence does not increase compliance in a field setting, although it does draw more attention than a human presence.

We believe that these results have particularly meaningful implications for the use of robots in places where ethical behavior of humans around them is critical, for example in government, educational, and healthcare environments.

In summary, during this four-year long project, Dr. Hoffman and his collaborators have advanced our understanding of human-robot interaction, and especially robotic nonverbal behavior, along the lines of team collaboration, robots listening to people’s disclosure, peripheral presence of robots in people’s environment, and their effect on ethical behaviors.

These findings can be used to build robots that are better suited for human interaction, paving the way to a shared human-robot labor economy, as well as for robotic products for people’s homes.