Robots for Kids - Understanding the Role of a SAR in HRI with Kids with ASD
This project consists of data coding and analysis of an interaction between robots and children with autism, using data collected by David Feil-Seifer. In this work we focus on understanding the role of the robot in the interaction, especially the ways in which the robot's role can be flexible in order to meet individual children's needs.
Enabling Continued Access to K-12 Education Using Telepresence
This work is motivated by a particular real-world challenge, enabling telepresence robots to provide continued access to K-12 education. Over 2 million children in America children have health or behavioral challenges that cause them to miss significant amounts of school each year. Being away from the school environment causes students to miss both educational and social experiences. Various solutions exist for addressing the educational experiences through home schooling, individualized tutoring, online learning, and other specialized instructional interventions, but the absence of cognitively stimulating peer-mediated educational experiences can have serious effects both in terms of social and cognitive development. Telepresence enables children to be remotely present at school and engage in regular classroom and social activities. This work focuses on understanding the needs of a telepresence system in the classroom and how it can be used for better learning and social outcomes.
Supporting Intergenerational Family Interactions in the Home
This project takes a novel approach to engineering desired group interactions and dynamics within a home with an older adult resident by using a socially assistive robot (SAR) as a member of the family team, embedded in a real-world environment and interacting with other family members in real time. Our focus in this project is on developing a SAR system capable of integrating into the family team with the goal of increasing group cohesion and providing a platform for positive family group interactions.
Robot as Moderator - Multi-Party Socially Assistive Robotics
Many types of multi-party interactions include a moderator, an agent that is responsible for directing the interaction, but is not necessarily directly participating in the task. In this work, the task of moderation is formalized as the process by which a goal-directed multi-party interaction is regulated via manipulation of interaction resources, including both physical resources, such as who is holding an object or a tool, and social resources, such as the conversational floor or participants' attention. The primary application domain for this model is in the domain of socially assistive robotics (SAR). We specifically focus on group learning with children, where the moderator ensures that several children working in a group interact successfully while completing a learning task.
Socially Assistive Robot Support of Habit Formation
In this project we are developing a framework that allows a Socially Assistive Robot (SAR) to support the process of habit formation. Habits are formed when cue-behavior pairs are reinforced with rewards over time when a person completes a specified behavior after encountering a cue for that behavior in the environment. We are exploring how a robot can support this process through providing reminders and social rewards for habits that promote health and wellness. We are developing this framework using an iterative process incorporating findings from social psychology and neuroscience into a computational formulation of SAR habit formation support.
Nonverbal Signaling for Human-Robot Collaboration
The goal of this research is to enable nonhumanoid robots to utilize nonverbal signals to improve human-robot collaboration. As robots start to not only share space, but directly interact with humans, they must convey information about their internal state. Such information promotes safety and enables coordination. We take inspiration from animation, psychology, psychophysics, and communication to develop new algorithms for generating state expressive cues using sound, light, and motion. We are applying this work to non-humanoid robots, such as the NASA Astrobee, which do not possess the communication modalities typically utilized in human nonverbal communication.
Enabling Computing Research in Socially Intelligent Human-Robot Interaction: A Community-Driven Modular Research Platform
Developing a low-cost robot platform for the Human-Robot Interaction (HRI) community. Assisting in organizing the 2015 AAAI Spring Symposium on "Enabling Computing Research in Socially Intelligent Human-Robot Interaction: A Community-Driven Modular Research Platform" and testing the software systems that will be deployed on the new platform.
Personalized Socially Assistive Robot Tutor of Teaching Number Concepts to Preschool Children
Robot tutors can provide effective supplemental support in the education of young children through delivering and engaging them in developmental materials and drills. However, no two children learn in exactly the same way. Personalization of instruction is a necessary component to reach a child’s full learning potential. It is an important and significant challenge to enable a robot tutor to customize its interaction and material to an individual child’s learning style. This work applies machine learning and statistical methods to learn personalized models of number concepts learning in preschool children, and leverage those models to personalize the robot tutor over multiple, repeated interactions.
Personalized Feedback for Reducing Anxiety in Hospitalized Children
This project is a collaboration between the Interaction Lab and Children's Hospital Los Angeles (CHLA) to reduce anxiety in children (aged 5-10) about to recieve an IV insertion. Working with Dr. Margaret Trost at CHLA, we are developing algorithms for personalized anxiety reduction feedback for use by a robot buddy (the Maki robot platform) that willinteract with children about to recieve an IV, along with a child life specialist. The study will be performed in three phases: 1) a preliminary data collection in which we observe child life specialists doing their jobs uninterrupted, test our sensing capabilities, and identify feedback strategies, 2) a second data collection using the Maki robot to give feedback based on observed child life feedback and sensed child emotional state verified in (1), in which we will match child responses to the robot's feedback with personality and pain anxiety measures, and 3) the full study, which will use personality/anxiety matching from pre-tests to determine the robot's feedback strategy. The full study will show the real-world application abilities of our personalization algorithms, although there will be no adaptation over time involved. Additional work will be done in a more controlled environment to test adaptation over time to what reduces anxiety best for each user.