Date: March 4, 2026
Speaker: Nathan Dennler, Postdoctoral Researcher, Massachusetts Institute of Technology
Title: Robot Optimization from User Interaction
Abstract: For robots to be effective in user-centered environments, they must adapt to the long-tailed distribution of user preferences. However, performing human-in-the-loop optimization is highly challenging: users cannot easily provide numerical values for their utility function, users have different notions of acceptable behavior, and asking users for repeated feedback can burden users. The first section of this talk introduces a representation learning approach that maps robot behaviors into a vector space compatible with existing optimization algorithms. The second section demonstrates that learning sets of feasible robot behaviors enables robots to adapt to users performing interactive physical exercises. The final section presents a human-in-the-loop optimization algorithm that efficiently learns a user’s desired robot behaviors from noisy feedback. By framing robot adaptation as human-in-the-loop optimization, we can develop robots that work well across a variety of environments, users, and tasks.
Bio: Nathan is a postdoctoral researcher at the MIT Computer Science & Artificial Intelligence Laboratory working with Prof. Andreea Bobu. Nathan’s research develops algorithms and methods that enable robots to efficiently adapt to users' preferences.
Nathan’s research has been recognized by RSS Pioneers, the ACM UIST Doctoral Consortium, and the University of Southern California Viterbi School of Engineering's Best Dissertation Committee. His work has been published in leading research venues, including Science Robotics, Transactions on Human-Robot Interaction, and the ACM/IEEE Conference on Human-Robot Interaction.
Previously, Nathan received a Ph.D. in Computer Science at the University of Southern California, where he was an NSF GRFP Fellow advised by Prof. Maja Matarić and Prof. Stefanos Nikolaidis. Nathan also holds a B.Eng. in Robotics Engineering and a B.S. in Computer Science from Worcester Polytechnic Institute.