Jan 10, 2018 | Atlanta, GA
Before new School of Interactive Computing Professor Seth Hutchinson came to Georgia Tech, before he made a bat robot fly, before he ever even delved into the worlds of robotics or artificial intelligence, he was asking one question:
Why am I supposed to do something a certain way? Why am I not supposed to do something else? What motivates people to make the decisions they make?
Growing up in an environment that promoted strict religious doctrine, he was, from an early age, interested in what motivated the rules that governed that community and what societal and psychological factors motivated differences elsewhere.
“I spent a lot of time trying to figure out where these ideas come from,” Hutchinson said. “As a kid, I spent a lot of time arguing these things, and I think that carried over into adulthood and informed my style of discourse now.
“Why? Why, why, why? Why do you think this should be true?”
Basically, Hutchinson, rather than being told and accepting something is true, wants to understand the layers that make it so. That desire for understanding led him on a path to studying artificial intelligence and robotics – a path that brought him to the School of Interactive Computing (IC) this year.
Understanding the scope of robotics
Hutchinson’s research started out in high-level artificial intelligence studies. It was one end of the robotics spectrum, which Hutchinson insists you must cover all of to really do robotics well.
“If you pick one narrow slice, it’s not long until you get frustrated because your robots don’t do anything,” he said. “They’re really good at planning, but can’t move or control their motions properly, and they don’t know how to process visual input. Or you get robots that look around and completely understand the world, but they don’t have algorithms that let them do anything about it.”
He began working with a Ph.D. advisor in computer vision in graduate school. Hutchinson classified it as an “accident” that he entered the field to begin with.
“I’m pretty sure my decision to work with him was motivated by nothing intellectual at all,” Hutchinson joked. “He just seemed like the coolest guy. I thought, ‘I want to be like him.’”
After working in that lab for a short time, a company called Cincinnati Milacron provided a large robotic arm. No one in the lab knew much about robots, so Hutchinson – with his drive to understand how things work – began to work with it.
This was in the 1980s, a time when, Hutchinson said, artificial intelligence was thriving. There was a community in AI, psychology, and even philosophy that was asking questions about what cognition means, how humans think about certain things, and how humans motivated to act.
“There was a real interaction between those communities and the robotics community,” Hutchinson said. “People making robots do stuff and those who were just trying to understand how one comes to decide to do something.”
It was that question again: Why?
“What is thought?” Hutchinson posed. “How does that work? Those are questions that have trapped young college students for years.”
Over the past three years, during which Hutchinson has been a professor at the University of Illinois, he and his team have done work on bat robots. It’s a project being funded under a grant from the National Science Foundation’s National Robotics Initiative. A colleague of Hutchinson’s had collaborated with others in the past who studied real bats, doing motion capture to analyze their flight in a wind tunnel, and the connection to robotics intrigued Hutchinson.
How do you design the folding mechanism on the wing to account for variations in aerodynamic equations, and how do you create a better model of aerodynamics to account for problems of weight or design?
“Basically, there’s a problem that needs to be solved,” he said. “You chop it into individual parts, and if you chop it the wrong way everything works too hard. Figuring out how to do that is very challenging, but very interesting.”
'I feel like I can do anything here'
That, more or less, is what ultimately drove him to Georgia Tech.
Hutchinson loved his time in Illinois and said that, on a daily basis when he roamed the campus, he would encounter some of the smartest people in the world. The problem? None of them spent much of their time thinking about robots.
For a long time, that was alright. Hutchinson could pursue robotics research without many other roboticists because there were so many theoretical problems that didn’t demand an experiment.
“Write down the equation, show it would work, and publish it,” Hutchinson said.
But that isn’t the case anymore. Hutchinson said it had become increasingly difficult to pursue real in-depth research – answering, or trying to answer, so many of those questions that had driven him throughout his life – without colleagues also doing robotics.
So, he made a list of about three or four schools that he would like to work at. Before he even had a chance to pursue any, he got a call from a friend at Georgia Tech.
“And, of course, Georgia Tech was on the list,” he said.
He came down for a visit and was blown away by the robotics researchers at the school.
“Mainly the young people,” he said. “I know there are fantastic robotics veterans here, but Georgia Tech has been so aggressive in hiring smart young people. I feel like I can do anything here. The pieces I don’t know, there are others who know those things.
“I anticipate having the most fun collaboratively I’ve had in my entire life. The spectrum of robotics is too big for just one or a couple of people. At Georgia Tech, there are so many people in so many departments to complement each other.”