Task-Oriented Koopman-Based Control with Contrastive Encoder
This work extends Koopman control to high-dimensional systems, addressing a bottleneck in robotics and control for tasks like real-world perception-based control.
The paper tackles the challenge of controlling high-dimensional, complex nonlinear systems by introducing a task-oriented Koopman-based control method that uses reinforcement learning and a contrastive encoder to learn latent embeddings, operators, and controllers, achieving successful application to pixel-based tasks and a real robot with lidar observations.
We present task-oriented Koopman-based control that utilizes end-to-end reinforcement learning and contrastive encoder to simultaneously learn the Koopman latent embedding, operator, and associated linear controller within an iterative loop. By prioritizing the task cost as the main objective for controller learning, we reduce the reliance of controller design on a well-identified model, which, for the first time to the best of our knowledge, extends Koopman control from low to high-dimensional, complex nonlinear systems, including pixel-based tasks and a real robot with lidar observations. Code and videos are available \href{https://sites.google.com/view/kpmlilatsupp/}{here}.