Human-Robot Interaction for Hosting Activities
We are developing a collaborative robot that can conversations with people and can perform and interpret physical gestures and movement during the interaction, thereby "engaging" the person.
Background & Objective: We are investigating engagement in human-robot interaction by studying human-human engagement, developing rules to apply to human-robot interaction, and applying those rules to a functional robot. We have focused on hosting settings, where the robot can provide information about the shared environment. Typical hosting settings include museums, stores and homes. In the home, the robot uses its knowledge of the home and its contents to direct people to items they need, it helps perform tasks, and it keeps track of the location of the people in the home. In museums and stores, it directs people to locations of interest and can demonstrate objects of interest.
Technical Discussion: Our robot collaborates with people to perform the hosting tasks of demonstrating itself or the iGlassware system developed at MERL. People interacting with the robot do not need any training to interact. In its current form, the robot interprets head gestures to determine if the person is attending to itself and to objects of importance in the environment. It recognizes and responds to the normal head nods that people make during a conversation to indicate agreement or understanding of what is being said. The robot is fully mobile using a Pioneer II mobile base. Over 100 people have completed demonstrations with the robot. These studies show that (1) people nod more often if the robot returns their nodes, (2) they find the robot's gestures more natural than an unmoving conversational partner, and (3) they direct their attention more to the robot than the unmoving conversational partner.
Outside Collaborations: We have collaborated with members of the MIT Vision Group (http://www.ai.mit.edu/projects/vip/projects.htm) to use the Watson vision system and support vector machine learning algorithms to interpret head nods of people interacting with the robot.
Contact: Joseph Katz
| Technical Reports: | |
| The Effect of Head-Nod Recognition in Human-Robot Conversation | |
| Contextual Recognition of Head Gestures | |
| Explorations in Engagement for Humans and Robots | |
| 'Man-Computer Symbiosis' Revisited: Achieving Natural Communication and Collaboration with Computers | |
| Explorations in Engagement for Humans and Robots | |
| Nodding in Conversations with a Robot | |
| A First Experiment in Engagement for Human-Robot Interaction in Hosting Activities | |
| Engagement by Looking: Behaviors for Robots When Collaborating with People | |
| Where to Look: A Study of Human-Robot Engagement | |
| The Role of Dialog in Human Robot Interaction | |
Technology Areas:
Off the Desktop Interaction and Display
Artificial Intelligence
Spoken Language Interfaces
Modification Date: September 12, 2007
