David Fridovich-Keil joins us as an assistant professor this fall. His research spans optimal control, dynamic game theory, learning for control and robot safety. While he has also worked on problems in distributed control, reinforcement learning and active search, he is currently investigating the role of dynamic game theory in multi-agent interactive settings such as traffic. Fridovich-Keil’s work also focuses on the interplay between machine learning and classical ideas from robust, adaptive and geometric control theory.
Fridovich-Keil received his doctorate from the University of California, Berkeley, where he developed some of the first efficient techniques for solving noncooperative, game-theoretic motion planning problems. During his graduate studies, Fridovich-Keil briefly worked at the self-driving car company Nuro. His postdoctoral research focused on exploiting computational parallelism in stochastic optimal control problems.
What attracted you to Texas ASE/EM?
I’ve heard wonderful things about living Austin, so it’s been on my radar for a while now. And since my research focuses on control theory and robotics, I was looking at schools where are I knew there were strong programs in those areas. I also knew former students who had attended and they all had good things to say about their experience. When I interviewed, it was all virtual due to the pandemic, so I was only able to meet people on Zoom, but I really enjoyed talking with everyone I did meet with, and left with a very positive impression. I’d say the last that thing is that UT and Austin felt very familiar to me. I grew up in Atlanta, so it’s still generally in the southern U.S. Basically, ASE/EM checked all the boxes for what I was looking for, and I am happy to join the department.
What do you enjoy most about your research?
I think the thing that I like most is that I find it very easy to straddle this divide between doing something that’s intellectually interesting and subtle, and at the same time also very, very practical. I originally became interested in multi-agent interaction and game theory while working at a car company. It really doesn’t get more practical than asking a question like, “Would I want to sit in this car?” and it’s very intuitive to think about the fact that almost everyone drives a car, and then put myself in a driving scenario. For example, thinking about how I might react to a certain type of oncoming car at an intersection, or whether I drive in a reasonable way. This area of research isn’t just purely experimental and it’s not truly theoretical – I think there are elements of both, and that excites me. (Image: Nash equilibrium of traffic game computed in real-time)
Tell us about your teaching philosophy.
I really like to provide project-based learning for both undergraduate and graduate courses. I especially think it’s important to include projects in graduate courses – something that students might be able to turn into a future research project — so I try to provide opportunities for that as well. The other thing I try to be aware of is if students are asking the same question multiple times, chances are, it means there is a better way to explain it. Also, being aware that students come into some courses with very different backgrounds and skill sets. Because of this, I think it’s a very good idea to design a course with different parts that will be familiar to a variety of students.
How do you like to spend your free time?
Outside! I like hiking, cycling and just walking around the neighborhood. A few years ago, I got into frisbee golf, so I was really excited to see how many courses there are in this area. Other than that, I like to read fiction and play acoustic guitar.
David Fridovich-Keil joins us as an assistant professor this fall. His research spans optimal control, dynamic game theory, learning for control and robot safety. While he has also worked on problems in distributed control, reinforcement learning and active search, he is currently investigating the role of dynamic game theory in multi-agent interactive settings such as traffic. Fridovich-Keil’s work also focuses on the interplay between machine learning and classical ideas from robust, adaptive and geometric control theory.
