Our goal is to develop and evaluate human-machine interaction techniques that influence the user to engage in wellness-promoting behaviors involving physical, cognitive, and social activity. We are developing methods that will use engagement and persuasion to encourage desired wellness-promoting behaviors through socially assistive human-machine interaction (HMI). Different types of technologies may serve as platforms for delivering wellness-related HMI, including smart phones, PDAs, computer-based interfaces, and intelligent robots. This work is focused on developing socially assistive systems (SAS) using peer-level HMI in which the machine serves in the role of a knowledgeable embodied agent capable of providing time-extended, sustained and engaging interaction. We validate our methods on computer-based and robot-based agents. In this way we will both validate the developed methods on multiple technology platforms and embodiments, and compare their relative effectiveness and acceptability by the intended user population.
This research focuses on two types of wellness-promoting human-machine interactions: 1) exercise sessions (cognitive and/or physical); and 2) socializing sessions. In the former, the SAS provides exercise monitoring, coaching, and motivation, while in the latter, the SAS provides social contact and friendly reminders and encouragement to comply with health regimens (taking medicine, being physically active) as well as healthy habits (calling family, meeting friends). The two types of interaction share the common underlying goal of influencing human behavior. We take a twofold approach to reaching that goal: 1) we use steering to learn how to affect the userís behavior in the short term during an interaction; and 2) we also use motivation to affect the userís longer-term behavior over the course of an interaction and help to maintain their engagement in the interaction.
This research focuses on the development, evaluation, and user testing of a Socially Assistive Robot (SAR) exercise coach designed to motivate and engage older adults in a seated aerobic exercise task. Our SAR system approach incorporates insights from psychology research into intrinsic motivation and contributes clear design principles for SAR-based therapeutic interventions. Through our user studies, we have examined several different aspects of wellness-promoting socially assistive systems, including a comparison of an embodied versus a non-embodied (computer-based) system, the role of praise and relational discourse, and the effect of varying user autonomy.
In this project, we extended our previous work with the elderly performing "chair exercises" guided by a socially assistive robot. The exercise scenario utilized a socially assistive robot to instruct, evaluate, and encourage users to perform simple arm gesture exercises. The scenario was one-on-one, allowing the robot to focus its attention on the single user in order to provide timely, accurate feedback, and to maximize the effectiveness of the exercise session for the user. In the set up, the user was seated in a chair in front of the robot; the user and robot faced each other. The developed probabilistic activity monitoring systems affords the robot the ability to track the user's arm movements; the use of the Kinect sensor (as opposed to a monocular camera) is an extension of our previous work, in that the robot and arm motion is no longer restricted to the sides of the body (i.e., non-planar)Automated Proxemic Behavior Recognition and Production
This research investigates proxemics in human-robot interaction (HRI). Proxemics is the study of the dynamic process by which people position themselves in face-to-face social encounters. This process is governed by sociocultural norms that, in effect, determine the overall sensory experience of each interacting participant. To facilitate situated and mobile HRI, this research seeks to develop functional and socially appropriate probabilistic computational models of proxemics for the purposes of both autonomous proxemic behavior recognition (of one or many people) and autonomous proxemic behavior production (by a sociable robot).
This work is supported by National Science Foundation (NSF) grant titled "Socially Assistive Human-Machine Interaction for Improved Compliance and Health Outcomes" - Award number: IIS-1117279.
Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.