Human-robot joint action: Coordinating attention, communication, and actions.

摘要

Successful integration of robots into human environments to support daily activities, such as assisting with household chores, promises to have positive impacts on people's quality of life. Such integration requires robots to serve, assist, and work with human users in a variety of joint actions. Joint action is a form of social interaction involving interaction partners who work cooperatively to coordinate attention, communication, and actions to achieve a shared goal. This dissertation seeks to design coordination mechanisms for robots to support human partners in such joint actions.;In a series of five studies, I explore how human-inspired coordination mechanisms can be realized for autonomous interactive robots. In Studies 1--3, I focus on designing gaze cues and gestures for robots to direct human partners' attention in supporting effective communication. In Study 4, I investigate how a robot might monitor a human partner's gaze cues to predict the partner's intent and how the robot can utilize the predicted intent to perform anticipatory actions to achieve efficient joint action with the partner. In Study 5, I explore how a robot can adapt to its partner's working pace by monitoring the partner's actions as well as task progress. Results from these studies show that the robots using the human-inspired coordination mechanisms can engage people in joint actions to elicit greater team performance (e.g., reduced idle time and task completion time) and enhanced user experience (e.g., positive perceptions of the robot and the interaction).;This dissertation contributes to enabling natural, effective joint actions between humans and robots. Throughout my investigation, I develop autonomous robot systems that manage real-time sensing, planning, and acting to support joint actions with humans. I explore innovative approaches to modeling, generating, and evaluating social behaviors for robots. Moreover, results from the studies provide insights to understanding human-human joint action and designing human-robot joint action. This dissertation further motivates future research toward integrating robots into everyday human environments to engage people in joint actions to create social, cognitive, and task benefits.