Abstract:

We would like for robots to be able to adaptively help people in their day-to-day lives, but the state-of-the-art in robot learning is typically either under-informed about the needs and abilities of actual users or is designed and tested in highly-controlled environments and interactions that fail to reflect real-world noise and complexity. In our work, we focus on identifying the real-world situations where current human-robot interaction (HRI) and robot learning algorithms fail, and developing new methods that enable robots to robustly learn to assist non-expert teachers under real-world noise and complexity. This includes using human-centered design to develop more realistic simulated teachers for early algorithm development, incorporating both teacher and environmental reward into state-of-the-art deep reinforcement learning algorithms, finding new ways to model and take advantage of rich-but-noisy human feedback, and designing novel models that enable robot-robot collaboration to improve detection of human attention. Finally, throughout all of this work, we seek to break down the artificial disciplinary divide between service robotics for non-disabled users and assistive robotics for users with disabilities, and insure that our robots treat all users as valued partners who are integrated into the social and physical environments in which they live their lives.

Bio:

Elaine Schaertl Short is the Clare Boothe Luce Assistant Professor of Computer Science at Tufts University. She completed her PhD under the supervision of Prof. Maja Matarić in the Department of Computer Science at the University of Southern California (USC). She received her MS in Computer Science from USC in 2012 and her BS in Computer Science from Yale University in 2010. From 2017-2019 she worked as a postdoctoral researcher in the Socially Intelligent Machines Lab at the University of Texas at Austin. At USC, she received numerous awards for her contributions to research, teaching, and service, including being one of very few PhD students to have received all three of the CS department Best TA, Best RA, and Service awards.

Elaine’s research seeks to improve the computational foundations of human-robot interaction by designing new algorithms that succeed in contexts where other algorithms’ assumptions frequently fail, such as in child-robot interaction, in minimally-supervised public space deployments, and in assistive interactions. As a disabled faculty member, Elaine is particularly passionate about disability rights in her service work. In addition to having recently joined the new AccessComputing Leadership Corps, she is the Communications Chair and Community Liaison of AccessSIGCHI, an advocacy group that works to increase the accessibility of the 24 SIGCHI conferences.

Schedule – EST (UTC-5)

6:45 – 7:00: Networking (in Zoom)

7:00 – 8:00: Presentation

8:00 – 8:30: Q & A