Task-Oriented Koopman-Based Control with Contrastive Encoder

Xubo Lyu, Hanyang Hu, Seth Siriya, Ye Pu, Mo Chen
Proceedings of The 7th Conference on Robot Learning, PMLR 229:93-105, 2023.

Abstract

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: https://sites.google.com/view/kpmlilatsupp/.

Cite this Paper

BibTeX
@InProceedings{pmlr-v229-lyu23a, title = {Task-Oriented Koopman-Based Control with Contrastive Encoder}, author = {Lyu, Xubo and Hu, Hanyang and Siriya, Seth and Pu, Ye and Chen, Mo}, booktitle = {Proceedings of The 7th Conference on Robot Learning}, pages = {93--105}, year = {2023}, editor = {Tan, Jie and Toussaint, Marc and Darvish, Kourosh}, volume = {229}, series = {Proceedings of Machine Learning Research}, month = {06--09 Nov}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v229/lyu23a/lyu23a.pdf}, url = {https://proceedings.mlr.press/v229/lyu23a.html}, abstract = {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: https://sites.google.com/view/kpmlilatsupp/.} }
Endnote
%0 Conference Paper %T Task-Oriented Koopman-Based Control with Contrastive Encoder %A Xubo Lyu %A Hanyang Hu %A Seth Siriya %A Ye Pu %A Mo Chen %B Proceedings of The 7th Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2023 %E Jie Tan %E Marc Toussaint %E Kourosh Darvish %F pmlr-v229-lyu23a %I PMLR %P 93--105 %U https://proceedings.mlr.press/v229/lyu23a.html %V 229 %X 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: https://sites.google.com/view/kpmlilatsupp/.
APA
Lyu, X., Hu, H., Siriya, S., Pu, Y. & Chen, M.. (2023). Task-Oriented Koopman-Based Control with Contrastive Encoder. Proceedings of The 7th Conference on Robot Learning, in Proceedings of Machine Learning Research 229:93-105 Available from https://proceedings.mlr.press/v229/lyu23a.html.

Related Material