Learn, Baby Bot, Learn
Josh de Leeuw ’08 and his URSI mentor Ken Livingston are interested in how autonomous robots learn. “Basically, we take an `infant’ robot, and we give it nothing but a way to learn. We want the robot to figure out how to move around on its own without bumping into things. So we let it loose to explore, and we watch it learn.”
Q. How did you decide to come to Vassar?
I was originally planning to apply to conservatories for music. Music was a big deal for me, and I was all set to go with auditions for conservatories. Then I had a crisis and wasn’t sure if I wanted to commit myself to that. Vassar seemed like a good choice because I could come in and explore. Both of my parents teach here. Sure, it was convenient, but that’s not why I ended up here.
I was initially interested in music and political science. My first semester I took history, political science, a music history class—a nice sampling. After I took my Intro to Cog Sci class second semester, I went ahead and declared the major. I was pretty hooked. Even after declaring, I’ve still been able to take a lot of different courses. I’ve taken classes in music and computer science. It’s easy to do a lot of different things here.
Q. What was it that appealed to you about cognitive science?
I looked at cognitive science while I was applying and it seemed very interesting. I was particularly interested in our ability to perceive music and why music has such a powerful effect and generates such an emotional reaction. I took the introductory course and loved it and then decided I wasn’t really interested in the music perception aspect anymore. I read a lot of the literature and started to understand what people were doing in the field of music perception, and it wasn’t as exciting to me as I thought it would be. So I ended up in robotics. I had always been interested in computer science and programming. I never took any classes in it before Vassar, but I’d always tinkered around as a hobby. I took a course called Perception and Action which is all about our ability to perceive and act in the world. It goes from the human level down to insects and robots too. There’s a big robotics component and a lab portion where you explore perception and action in robots. That got me hooked.
Q. Did you ever think you’d end up doing what you’re doing now?
No, but looking back, it makes sense. I’ve always been interested in hands-on exploration and building. There’s a creative aspect to it. I’m happiest when I’m being creative, and this is a great creative outlet. And there’s engineering involved, too. You can do it at so many different levels. You don’t need to learn to run the electronics to just dive right in and go at it from a cognitive perspective. It’s a creative process, but also scientific.
Q. Have you done URSI?
I’ve done URSI for two summers, and I’m going to do it again this summer. My first summer was after my sophomore year. At that point, I had only taken one course that had to do with robotics, and I’d taken some computer science, but not really a whole lot. I could comfortably program the robot, but I didn’t have much of the other background. But that was okay because the whole URSI project is looking at how robots can learn in a more similar way to how humans learn. The computer and the brain are better at different things. We want to model what happens in the human brain in a computer. We think that makes sense.
Q. What kinds of things do you have the robots do?
We explore basic navigation, which is a pretty difficult task. Basically, we take an `infant’ robot, and we give it nothing but a way to learn. We want the robot to figure out how to move around on its own without bumping into things. So we let it loose to explore, and we watch it learn. We record how it learns, what improves the learning and what doesn’t. I’ve given about four talks on it now, two at URSI and then I went to a conference at Rutgers this fall and met other people interested in this. I presented a poster, and part of that was giving a talk on what we had done. Recently, I gave another talk at West Point as part of a Hudson Valley Academic Engineering Forum. I talked about my thesis which is an extension of what I’d been working on before.
We actually have a new model now that we don’t think anyone else has done before. So part of the URSI project this summer will be to transfer my thesis project into a journal article. Hopefully we’ll get it published.
Q. So this has really clicked for you.
It’s been great. I don’t think there could have been a better match. As an undergraduate, I was able to do any research that I wanted to on my thesis this year and have access to all the equipment in the lab that I needed. Most grad students don’t get to touch that equipment right away, and they have a faculty member who’s steering them and deciding what the research direction should be. That was part of the process here, but by senior year, I was able to go off and do what I wanted to do. I’ve been working with Ken Livingston for the most part. Most of this was his idea, but he let me set the direction for my thesis project. I was able to test my own ideas to see if they worked. And it turns out that they do. So we’re going to keep exploring this over the summer.
Q. Have you been involved in the robot competitions?
I competed my sophomore and junior years. The competition is part of an independent study class. There are several teams of three or four students who compete in a game that’s sort of a variation on capture the flag. I had great teams, and we won both years. You’re given a set of rules and a budget—the rest is up to you. So you put five to 10 hours a week in on the robot and more as you get closer to the competition. It’s nice to be able to spend that kind of time on an independent project.
Q. What did you win?
A nice trophy, bragging rights!
Q. What else are you involved in?
This past year I was chair of the judicial board, which is an elected position. When students are accused of violating college regulations, they take their case before the panel. I was sort of guiding the students. It was important to get involved. It gives you perspective and shows that there are two sides to every story. Nothing is clear cut. It’s just like science.
Q. What are your post graduation plans?
I’m doing URSI. And then, it’s up in the air. One of my professors has applied for a grant, and if he gets it, there’s a good possibility that I’ll have a full-time research position next year. Beyond that, I haven’t decided what I’m going to do. This field is taking off, so I could get a master’s and work commercially. But I’m also interested in getting my PhD, teaching at the university level, and doing research.