Skip to main content
The Abstract

Dogs, Math and Computers: How One Researcher Gets His Ideas

David Roberts and one of his collaborators.

David Roberts smiles a lot. And why not? He’s a computer scientist whose research ranges from artificial intelligence and video games to developing new technologies that better connect dogs and humans. He loves his work, and lots of people are interested in it.

But what inspires the research in the first place? I know that researchers need expertise, an understanding of the scientific process and experimental design (and patience, of course). But you also need an idea to test in the first place. Where do those ideas come from?

I decided to ask people, starting with Roberts.

For Roberts, the ideas started at home, when he was still a kid.

“I come from a very mathematical background,” Roberts says. “Both of my parents are very successful professors in math departments at big universities. My dad, in particular, is an applied mathematician whose specialty is mathematical models of social and environmental phenomena, which means I had tons of exposure when I was growing up to explanations of things I found interesting using mathematical arguments. One of my favorite examples is the ‘One-Way Street Problem’ in graph theory where the question of how to direct traffic flow through a street network can be examined formally.”

One of David Roberts's collaborators gives him some positive reinforcement.
One of David Roberts’s collaborators gives him some positive reinforcement.

I could see the jump from mathematics to computer science, but how did dogs become part of his research area? Well, for starters, Roberts is a dog lover and amateur dog trainer. The secret ingredients, however, may well have been grad school and online chat rooms.

“When I was a graduate student, and had more free time, I was very active on an internet forum dedicated to the native Japanese hunting dog breeds, such as the Akita Inu, Shiba Inu, or Kai Ken,” Roberts says.

“Many of the forum’s most active members were extremely smart when it came to dog training, and (like me) enjoyed debate. Before long, I gained a reputation for using ‘math lessons’ to win arguments. For example, I used math (combinatorics, specifically – I won’t go into it here) to explain why managing the interactions of more dogs in the face of limited resources (like a prized toy) is exponentially more complicated with each additional dog.

“One day, it struck me that my arguments weren’t solely an intellectual exercise and I began to actively think about how to leverage my knowledge of dog training and behavior to develop new algorithms and technologies. When I came to NC State, the first person I reached out to as a potential collaborator was Barbara Sherman, an animal behavior expert in the veterinary school. Then I met [electrical engineering researcher] Alper Bozkurt during new faculty orientation shortly thereafter. And the rest is history.” [Note: Roberts, Sherman, and Bozkurt are co-PIs on an NSF grant to develop technology for computer-assisted dog training – which was used to develop the disaster-response technology for dogs.]

So, where do Roberts’s ideas come from?

“For me, and I fully acknowledge that I may not be normal, the chain is: conversation, then idea, then more conversation, then research,” Roberts says. “I approach my profession with the mantra that ‘you never know where your next idea will come from,’ so I try to stay as well rounded and informed about as many things as I possibly can.

“While I may not be the absolute smartest or most sophisticated researcher, I pride myself on my ability to see connections and relationships between things that may appear at face value to be very different. So, in that vein, my ideas typically come to me in one of three ways: they spring up when I let my mind wander during a conversation; I wake up in the morning with an idea I didn’t have before going to bed the night before (typically inspired by a conversation I had, article I read, or show I watched the day before); or they come to me while lost in thought doing a relatively mindless task, like walking my dogs or taking a shower. From there, I seek out conversations with people I perceive as more knowledgeable than me in the relevant topics. The research arises when I’m able to communicate the ideas effectively to those people, and they get as excited as I am.”