Inspired by the problem of understanding the interaction and dynamics of robot swarms from observations, I spent my first year as a PhD student investigating different methods for modeling dynamical systems from data. Hence, I researched largely in the field of data-driven system identification, more specifically with deep learning techniques.
At the beginning of my second year, I concluded my work with a paper entitled "Learning Nonlinear Dynamics and Chaos: A Universal Framework for Knowledge-Based System Identification and Prediction",
which presents a neural-network-based framework for learning general classes of dynamical systems. I successfully demonstrated learning stiff systems, spatiotemporally chaotic systems, and noisy systems using my framework.
In the future, I plan to expand on this framework for applications in data-driven control, adaptive sampling of dynamic environments, as well as field robotics.