Last summer, I worked at the Flight-Systems and Controls Lab (FSC) at the University of Toronto Institute for Aerospace Studies (UTIAS).
1. the research
From May to August 2025, I worked with and under MSc student Kevin Chen in learning, assisting and expanding his work on nonlinear control and stability, specifically exploring contraction metrics for quadrotor systems. The core challenge was guaranteeing convergence without relying on local linearization.
In late August, I was able to join Mr. Kevin Chen and Prof. Liu on site, and provided assistance in Kevin's neural network that generates valid contraction metric matrices for his quadrotor system. I also began attempting to recreate his work with a much simpler non-linear system, but with a different method of network training. That is, I defined my loss function based upon its adherence to a contraction condition derived from the system's Jacobian, rather than the method of using a training set of pre-computed contraction metrics for various states.
2. the culture
Research wasn’t confined to the lab. A surprising amount of thinking happened away from equations entirely. As my work was part of the Aerial Robotics program, I worked alongside other summer students exploring different projects. With a background in rocketry and a growing interest in flight systems, I focused on trajectory tracking while making time for lunch and the games around the lab.
There was an obscene amount of ping pong, but chess was the main event. One of my goals that summer was to increase my rating, so the games became an unexpectedly serious part of the routine.
After removing time constraints and a few games of what would more accurately be described correspondence chess, I eventually snagged a win after a quick-move blunder! It was a satisfying way to cap off the summer, and I look forward to continuing both my research and hobby chess in the academic year ahead.
3. future work
The summer was a launching point. I look forward to continue exploring nonlinear control, perhaps integrating machine learning techniques to adaptively learn contraction metrics in real-time. Currently, I continue to work with Mr. Kevin Chen through hybrid meetings throughout my academic year, and am working on an extension to his research on CCM loops for quadrotor with slung payload systems to 360° roll-maneuvers for racing drones. To end the session, I was invited to Prof. Liu's internal 'LiuTClub' and have since benefitted greatly from his mentorship. I was also asked to return as the sole undergraduate mentor for the next summer cohort at FSC UTIAS, which I look forward to. Overall, my experience here has solidified my passion for intersections of machine intelligence and control theory, and I look forward to where this path takes me next.
— ashlyn lee · end of file.