Luke Strgar

I am a PhD student in Computer Science at Northwestern University, advised by Sam Kriegman in the Xenobot Lab. I have also spent time as a student researcher on the Paradigms of Intelligence team at Google.

I received my MSc. in Computer Science from the University of Texas at Austin, where I worked with David Harwath on self-supervised speech segmentation. Before that, I earned my B.A. in Computer Science from the University of California at Berkeley, working with Justin Remais on mathematical models of infectious disease dynamics. Between these degrees, I held research positions with Tim Gardner at the University of Oregon and Young Hwang Chang at the Knight Cancer Institute.

Research

I am interested in intelligent systems that adapt across multiple timescales. Biological organisms not only evolve over generations, but also learn extensively during their lifetimes; these intertwined processes shape their bodies, behaviors, and capabilities. To study this interplay, I develop algorithms and simulations that jointly design robot morphologies and control policies, using both evolutionary and gradient-based methods, to produce agents capable of adaptive behavior. A central focus of my work is understanding how to induce properties such as modularity, compositionality, and diversity, which are widespread in natural organisms but challenging to reproduce in artificial systems.

Selected Publications

Accelerated co-design of robots through morphological pretraining
Luke Strgar, Sam Kriegman
Preprint, 2025
Evolution and learning in differentiable robots
Luke Strgar, David Matthews, Tyler Hummer, Sam Kriegman
Robotics: Science and Systems, 2024
Phoneme segmentation using self-supervised speech models
Luke Strgar, David Harwath
IEEE Spoken Language Technology Workshop, 2023

News

February 2025
I started work as a student researcher on the Paradigms of Intelligence team at Google in Zurich, Switzerland
July 2024
Watch my submission to the Artificial Life Virtual Creatures Competition here
March 2024
Evolution and learning in differentiable robots was accepted to Robotics: Science and Systems