Deep Learning of Koopman Representation for Control
This work addresses control design for dynamical systems, but it appears incremental as it applies an existing method to new data without broad SOTA claims.
The authors tackled the problem of optimal control for dynamical systems by developing a data-driven, model-free approach using deep neural networks to learn the Koopman operator, and demonstrated its capability on two classic systems in the OpenAI Gym environment.
We develop a data-driven, model-free approach for the optimal control of the dynamical system. The proposed approach relies on the Deep Neural Network (DNN) based learning of Koopman operator for the purpose of control. In particular, DNN is employed for the data-driven identification of basis function used in the linear lifting of nonlinear control system dynamics. The controller synthesis is purely data-driven and does not rely on a priori domain knowledge. The OpenAI Gym environment, employed for Reinforcement Learning-based control design, is used for data generation and learning of Koopman operator in control setting. The method is applied to two classic dynamical systems on OpenAI Gym environment to demonstrate the capability.