Computationally Efficient Data-Driven Discovery and Linear Representation of Nonlinear Systems For Control
This work addresses the challenge of controlling nonlinear systems more efficiently and accurately for applications in robotics or engineering, but it appears incremental as it builds on existing Koopman operator theory and deep learning methods.
The paper tackles the problem of system identification and linearization of nonlinear systems for control by proposing a data-driven deep learning framework with recursive learning based on Koopman operator theory, resulting in a method that is trained more efficiently and is more accurate than an autoencoder baseline, as demonstrated in simulations on a pendulum system with noisy data.
This work focuses on developing a data-driven framework using Koopman operator theory for system identification and linearization of nonlinear systems for control. Our proposed method presents a deep learning framework with recursive learning. The resulting linear system is controlled using a linear quadratic control. An illustrative example using a pendulum system is presented with simulations on noisy data. We show that our proposed method is trained more efficiently and is more accurate than an autoencoder baseline.