CAREER: Designing Socially Intelligent Robots that Collaborate with Groups of People

University Of Chicago, Chicago IL

Investigators

Abstract

Social robots can help people in a variety of ways. They can serve as tutors to elementary school students. Social robots can contribute as teammates in collaborative human-robot teams. They can also provide daily care to older adults. These social robots have potential to improve the lives of everyday people. Yet, social robots will fail to reach their full potential unless they have social skills. It is especially important that a robot’s social skills allow them to interact with groups of people. If they do not have these social skills, people will ignore them. Robots without social skills will also interrupt people and support harmful biases. If robots have social skills, they can have effective communication in group settings. They can act with social awareness, and better encourage team collaboration. In this project, the research team will design three robot social skills. These skills will enable successful interactions between a robot and a group of people. These social skills first include building rapport with people in a group. Second, they will adapt to the human-human relationships present in the group. Last, the robot will promote the group's social norms. First, the research team will develop and test each of these social skills. Next, they will combine these three social skills into one robot. This robot will interact with groups of 3rd-5th grade elementary school students. The robot will help these students with a collaborative learning activity. The research team will study the benefits of the robot’s social skills. This research will enhance human-robot communications, group dynamics, and group performance. The robot social skill designs created can be adapted for robots in other contexts. These social skills may apply to search and rescue as well as hospital delivery robots. These skills will also help to improve human-robot group outcomes. The goal of this project is to design social skills for a robot to use to collaborate with groups of people. Recent advancements in computing have provided robots with increased technical abilities. Yet, it will be difficult for robots to integrate into daily life without appropriate social skills. Having usable social skills will enable improved interactions with groups of people. The research team will design three robot social skills that will enable robots to better collaborate with groups of people. The first skill is building rapport with groups of people. The second set of skills is adapting to human-human relationships. The third skill is promoting group norms. To develop these social skills, the research team will rely on prior work in psychology. They will use data-driven techniques to define a set of constituent fine-grained robot behaviors. These social skills may include engaging in mutual gaze when making key decisions. Another social skill may be learning to ask appropriate follow-up questions. The design and implementation of these social skills will be adaptable to different robots and contexts. The research team will perform laboratory studies with groups of adults. These studies will determine the effectiveness of the developed robot social skills. They will explore if the social skills have the intended effect and influence group outcomes. The final step for this research is a real-world evaluation. The research team will adapt the social skills developed in earlier project phases. These adapted skills will enable a robot to assist with collaborative learning. The research team will perform this study with groups of 3rd-5th grade students in elementary school classrooms. In this evaluation, the aim is to show the positive influence of the robot's social skills on both student engagement and learning outcomes with collaborative learning tasks. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.