Research

Our research is in human-robot interaction at the intersection of assistive robotics and robot learning. The goal of our work is to make data-driven learning robots more responsive to and supportive of users, especially disabled users. From the assistive technology side, this includes developing next-generation disability technologies that focus on holistic well-being, including joy and thriving, consider disabled people in their social and community contexts, operationalize theoretical models of disability other than the medical or charity models, support multiply-, contextually-, and intermittently-disabled people as well as people with well-defined and static needs, and address accessibility throughout user-facing technologies rather than only in dedicated assistive technologies. From a general robotics perspective, this involves eliminating the artificial separation between service and assistive robotics, operationalizing long-standing insights from the disability community to create robotic systems that are collaborative with users, supportive of user agency, responsive to user needs, and that model users as intelligent, agentic, and helpful. Our work can be grouped into three themes: robot learning in groups and crowds, robot learning from human teachers for on-the-fly adaptation, and robot learning when users are creative and empowered.

To get a sense of our recent work, you can browse the AABL Lab's publications page.

Research Areas

Fluent Interaction with Groups and Crowds

This work allows robots to quickly and naturally influence, understand, and learn from people in groups and crowds. Our goals it to make robots responsive and appealing social partners, while also accomplishing instrumental tasks relating to their embodiment (e.g., driving around a building to show a visitor where to go, or picking up objects to clear them off a table). Key research problems include:

  • Modeling group interactions and making decisions about when and how a robot should intervene.
  • Understanding how to learn from diverse users in public spaces.
  • Designing learning algorithms that are robust to the ways that humans teach and interact with robots when other people are around.

Learning from Users for On-the-Fly Adaptation

General-purpose robots are unlikely to be deployed with all of the knowledge they need to complete every possible task. We expect that robots will need to learn after deployment, whether to customize their behavior to an individual user, or to learn to handle situations that were not anticipated by the robot designers. In particular, users are a rich source of information for robots, but the data obtained from these interaction is often noisy and sparse. Our work in this area both enables robots to learn more effectively from non-expert users in noisy real-world environments and equips robots with the social and interaction skills to help non-expert users provide more useful examples. Key research problems include:

  • Designing new types of interaction that result in better teaching from non-expert users, including improving their mental models of the robot's needs, providing more natural interfaces for teaching, and allowing the robot to actively request helpful examples.
  • Modifying existing learning algorithms with knowledge about common errors made my non-expert users in order to make them more robust to real-world teaching.
  • Developing new active learning algorithms that enable robots to participate in the learning process and acquire the most useful information from human teachers.

Empowering Diverse Users

Finally, our work is focused on understanding the needs of a wide range of users, especially disabled users. We are interested in how learning and interaction changes when users are treated as agentic, creative, and wise problem-solvers. We use inclusive design and participatory research approaches to gain a better understanding of the needs and skills of these users and develop new learning algorithms that empower them. Key research problems include:

  • Developing interactions that support user autonomy during robot learning and deployment
  • Applying human-centered and participatory design methods to the development of robots that learn from users
  • Leveraging practical and ethical insights from disability studies and the disability community to understand and improve assistance for both disabled and non-disabled users.