Session: 20-17-01: Rising Stars of Mechanical Engineering

Paper Number: 172168

Closing the Loop Between Learning and Communication for Assistive Robot Arms 

Research funded by this Faculty Early Career Development Program (CAREER) award intends to formalize bidirectional exchange between humans and robots as a closed-loop dynamical system. The research is inspired by the need for assistive robots that help elderly and disabled adults perform everyday tasks. The crucial challenge lies in fostering seamless and safe interaction between humans and robots. When people rely on assistive robots, they need robots that learn what the user wants (i.e., the robot understands the human) and clearly communicate their intent (i.e., the human understands the robot). The investigator's team will develop hardware and software tools to facilitate communication between robots and humans through visual, auditory, and haptic interfaces. This project will also support robotics education by i) teaching deaf and blind high school students to build interfaces that communicate with assistive robot arms; ii) mentoring college seniors as they program "learn & share" robots that test the limits of the unified framework; and by iii) bringing engineering faculty outside of traditional inclusion groups into a safe space to discuss their journeys.

The insight driving this research is that effective communication informs learning and vice versa. Users with knowledge about their robot's learning enhance their ability to operate it, while the robot's learning dictates what it should communicate. With this insight in mind, research performed in association with this CAREER award has three main objectives: (from human to robot) learning to assist the human robustly and intuitively, (from robot to human) communicating robot learning with personalized and interpretable feedback, and (co-adaptation) unifying learning and communication under a single, interconnected dynamical system that allows the robot and the human to interact and adapt. User studies will evaluate these contributions, involving participants controlling a robot arm in daily living activities. A unified formalism where learning and communication are two components of the same mathematical framework is the main scientific contribution. By combining these elements, the project transcends the limitations of separate development tracks for learning and communication, advancing knowledge in previously unattainable ways. The project's educational and outreach plan complements these research objectives and teaches the next generation of engineers to deeply consider how users understand and co-adapt to new autonomous technologies.

Our recent work towards this award includes research at the intersection of stability and imitation learning, research on learning from human videos, and research that develops assistive eating devices for adults living with mobility limitations. My poster will present these recent results and highlight my group's progress towards robots that learn from humans --- while simultaneously making it clear what they have learned.

Presenting Author: Dylan Losey Virginia Tech

Presenting Author Biography: Dylan Losey is an assistant professor in the Department of Mechanical Engineering at Virginia Tech. His research interests lie at the intersection of human-robot interaction, robot learning, and control theory. Specifically, he develops collaborative robots that understand --- and are understood by --- their human partners. Dylan was previously a postdoctoral scholar at Stanford University. He earned his doctoral degree in Mechanical Engineering from Rice University in 2018 and his bachelor’s degree in Mechanical Engineering from Vanderbilt University in 2014. Dylan is an NSF CAREER recipient and has won best paper awards from the Conference on Robot Learning, the IEEE/ASME Transactions on Mechatronics, and the IEEE Transactions on Haptics.

Authors: