‘People are sometimes unpredictable, that’s what makes this research fun!’

11 Oct 2018

Henny Admoni is an assistant professor at the Robotics Institute at Carnegie Mellon University. Image: CMU

Henny Admoni of Carnegie Mellon University wants to build the next generation of assistive care robots, but she’s not trying to make them ‘smart’.

With a rapidly global population, assistive robots are set to become increasingly common in the years and decades to come, particularly in care homes for the elderly.

One researcher looking to bring these robots closer in line with what humans want is Henny Admoni, an assistant professor at the Robotics Institute at Carnegie Mellon University (CMU).

After completing her PhD in computer science in 2016 at Yale University, she went on to work at the university’s Social Robotics Lab. She then undertook a postdoc at CMU’s Personal Robotics Lab, and last year joined the Robotics Institute in her current role.

What inspired you to become a researcher?

I was very lucky to be part of a science research course in my high school. That was the beginning of my love for research; doing high school science research projects and entering them in local science fairs.

The summer before my senior year of high school, I interned in Prof Denis Pelli’s psychology lab at New York University, and that experience really solidified my interest in pursuing academic research.

I went all the way to the Intel International Science Fair with the project from that summer, and I also learned a lot about how university research labs worked.

Can you tell us about the research you’re currently working on?

My current research investigates how assistive robots can use human behaviour to predict when people need help, and then figure out the right action to take. For example, one robot I work with is an assistive robot arm (a commercial product manufactured by Kinova Robotics). This arm helps people who have a motor impairment to do everyday tasks like picking up a glass of water to take a drink.

My research is on how subtle human cues – like where people are looking – can help us predict what task they will do next, when they’re having trouble controlling the robot or whether they’re uncertain about their goals. In other words, I want the robot to be able to read social cues, just the way people do, to be the most effective assistant ever.

Another angle for the same research is with socially assistive robots, like robot tutors or therapy assistants. In that case, being able to read subtle, nonverbal cues from people will make the robots better at figuring out what information to convey or when someone is feeling overwhelmed – just the way the best teachers do. The goal is never to replace human interaction, it’s to augment what people get from human-human interactions by adding personalised human-robot interactions.

In your opinion, why is your research important?

The goal of my research is to use robotic technology to improve people’s quality of life. In the physically assistive case, for example, I hope that reading human behaviour to improve robot assistance will help make it so that an assistive robot requires less physical and cognitive effort to operate.

In the socially assistive case, I hope that robots can fill a gap as one-on-one partners that help people practise skills they’ve learned from their teachers.

What commercial applications do you foresee for your research?

Most of my research is still at the fundamental science stage; we’re still developing an understanding of how people use robots, how their behaviours correspond to their needs and how to incorporate that information into robot actions.

The path to commercialisation would be to put the algorithms and technologies my lab is developing into existing assistive robots – like the Kinova robot we use – to make those robots easier for people to operate.

What are some of the biggest challenges you face as a researcher in your field?

The biggest challenge – but also the most interesting part of my research – is that we don’t know a lot about how people interact with these kinds of assistive robots.

The field of human-robot interaction has only been around for about two decades and, though we’re building on many more years of psychology research, there are new insights to gain when we look at how people act with and around robots. People are sometimes unpredictable, and that’s what makes this research fun!

Are there any common misconceptions about this area of research?

A major misconception is that intelligent robots are going to reach a point where they outsmart humans. Roboticists use the word ‘intelligent’ to mean a very specific concept: robots that can autonomously perceive the world, process that information and take action, without needing a human to control them. But robots aren’t ‘smart’ in the way people are, nor do I think we’re going to reach that point in my lifetime, which is good, because that’s not our goal.

Our goal is to understand people well enough to design technologies that help them, that enable them to live more fulfilling lives, including enabling them to socially interact with other people.

What are some of the areas of research you’d like to see tackled in the years ahead?

As robotic hardware and software improves, we’re seeing companies try to get personal robots into homes. Some of them, such as Jibo and Mayfield Robotics, started to sell personal robots but eventually reduced operations or closed. I’d love to see a deeper investigation of how people use robots in their homes, and how we can make home robots a viable technology for people.

A lot of human-robot interaction research to date has, necessarily, involved lab-based studies. Doing studies in the lab allows us to control the environment, including the robot, but it leads to somewhat synthetic interactions. I’d love to see more human-robot studies out in the world.

As more robot technology becomes commercially available – whether in the form of autonomous vehicles, intelligent personal assistants or assistive robots for healthcare – we researchers have an easier way to investigate more naturalistic human-robot interactions. I’m very excited about that trend.

Are you a researcher with an interesting project to share? Let us know by emailing editorial@siliconrepublic.com with the subject line ‘Science Uncovered’.

Updated, 4.02pm, 12 October 2018: This article was amended to provide more up-to-date information with regard to the areas of research Henny Admoni would like to see tackled.