Kyle Morgenstein

 I'm excited about building intelligent social robots.

Hi! My name is Kyle Morgenstein and I am a fourth year PhD student in the Human Centered Robotics Lab at the University of Texas at Austin advised by Luis Sentis. I am also a researcher at the Boston Dynamics AI Institute on the Ultra-Mobility Vehicles (UMV) Reinforcement Learning (RL) team advised by Gabe Nelson and Farbod Farshidian

I'm excited about human-centered design for social robots and the potential for robots to radically improve the lives of humans. My work seeks to answer questions in reinforcement and representation learning surrounding emergent social behaviors for robots that work near and interact with people. More specifically, I employ learning-based sensor fusion techniques to give robots the ability to "feel" contacts to proactively reason about external stimuli and exploit environmental dynamics. 

Previously I graduated from the Massachusetts Institute of Technology (MIT) in 2020 with degrees in Aerospace Engineering and Earth Atmospheric & Planetary Science, where I won the Christopher Goetze Award for undergraduate research. My previous research was primarily in computer vision and controls at NASA JPL, Boeing, and MIT Lincoln Labs.

When I'm not playing with robots, I am an an avid hiker, rower, and tinkerer. I also enjoy cooking, drawing, lifting weights, chocolate ice cream (edit: I am now lactose intolerant), and Club Penguin.

Current Research

Learning Contact-based Navigation in Crowds

What social norms can robots learn from humans in order to behave in ways that are safe and trust-promoting? This research challenges the current use of simple human simulation models and contact-forbidden control paradigms and explores how estimating the human’s perception of the robot may produce more socially accepted behavior. See project here.

Information-theoretic Multi-agent Search

I am applying multi-agent rienforcement learning methods to the problem of dynamic object search / active sensing motivated by traditional methods in information-theoretic optimal area coverage search. 

For my undergraduate senior thesis, I experimented with filamentous Cyanobacteria from Shark Bay, Australia to understand how silica controls in the Proterozoic ocean affected the preservation potential of ancient microbial mats. See a talk I gave for the United Nations Office of Outer Space Affairs here.