skip to main content

The robot is in

Computer science and engineering researcher develops socially assistive robots for use in health care settings

David Feil-Seifer

David Feil-Seifer's interest in robotics started with a canceled robotics class during his undergraduate days at the University of Rochester.

"A group of five of us, we just got really mad about it," he laughed. "So we formed a team to enter a competition. We ended up winning the competition, but we spent most of our junior and senior years on this, and in the middle of that I realized I would rather do this than anything else."

A couple graduate degrees later, Feil-Seifer is a new assistant professor in the Department of Computer Science and Engineering, and he's applying his love of robotics to the health care arena.

Feil-Seifer works in the emerging area of socially assistive robotics, which focuses on developing robots that can help people through social interaction rather than physical intervention.

"These are robots that are designed to coach you on a physical therapy regimen, act as a companion, or encourage you to make positive health choices," he said. "And I'm really interested in this topic because I believe there are a lot of areas in the current health care system where professionals can be provided assistance."

Feil-Seifer sees the main applications for his work, at least initially, in the ability for robots to monitor repetitive or tedious tasks. For example, post-stroke recovery therapy that requires repetitive, small exercises for six or more hours a day.

"That's a great task, it's something where it's a direct benefit, it's tedious for everyone involved, and a robot has patience as long as it has batteries," he said. "But this robot is going to be with you for 30 hours a week - it had better stay interesting. Otherwise you're going to tune it out."

In fact, the main challenge Feil-Seifer is currently tackling is how to design a robot that is interesting and engaging to a person over the course of sustained interaction.

"If a robot or a computer or any autonomous system starts giving you the same piece of feedback over and over and over again, you're going to start to tune it out," he said. "We need a robot or a computer or a web service to be able to take the content that they want to express, know how many times that's been delivered before and then change its behavior."

In order to design a robot capable of that kind of subtlety in interaction, Feil-Seifer is bridging a lot of disciplinary divides to draw on research in human interaction from fields such as psychology, anthropology and philosophy.

"I've worked with professionals in all of those fields, and I plan on continuing to," he said. "As I get started, I'm actively looking for collaborators."

In addition to all the research problems that come with designing robots to interact with humans, Feil-Seifer also specializes in how robots can be used with kids, which poses its own set of challenges. One early lesson came while he was in graduate school when he introduced a bald, gray robot with a female voice to a group of children.

"Kids were really entertained by it," he said. "But each child came up to me at the end of it and said, ‘why is the boy robot talking with a girl's voice?' Some of these kids found it really off-putting."

In addition to meeting initial expectations, Feil-Seifer hopes to develop socially assistive robots that can adapt over time, much like a human would.

"The first time you step into a physical therapist's office, usually they're trying to establish a friendly relationship," he said. "But as you get into the later sessions, they use that rapport to communicate more efficiently. Looking at that change in personality over time is something we're very interested in."

As Feil-Seifer gets his research underway at the University of Nevada, Reno, he's focusing on getting robots and starting initial testing with college students. But he's got his sights set on big goals down the road.

"What I hope to do in the future for an advanced class is actually identify client community and then have people try and design robots to address their needs," he said. "That is going to be a couple years down the road, but it's something I'm trying to build toward."

Take the next step...