Action-Conditional Recurrent Kalman Networks For Forward and Inverse Dynamics Learning

Vaisakh Shaj, Philipp Becker, Dieter Büchler, Harit Pandya, Niels van Duijkeren, C. James Taylor, Marc Hanheide, Gerhard Neumann
Proceedings of the 2020 Conference on Robot Learning, PMLR 155:765-781, 2021.

Abstract

Estimating accurate forward and inverse dynamics models is a crucial component of model-based control for sophisticated robots such as robots driven by hydraulics, artificial muscles, or robots dealing with different contact situations. Analytic models to such processes are often unavailable or inaccurate due to complex hysteresis effects, unmodelled friction and stiction phenomena, and unknown effects during contact situations. A promising approach is to obtain spatio-temporal models in a data-driven way using recurrent neural networks, as they can overcome those issues. However, such models often do not meet accuracy demands sufficiently, degenerate in performance for the required high sampling frequencies and cannot provide uncertainty estimates. We adopt a recent probabilistic recurrent neural network architecture, called Recurrent Kalman Networks (RKNs), to model learning by conditioning its transition dynamics on the control actions. RKNs outperform standard recurrent networks such as LSTMs on many state estimation tasks.Inspired by Kalman filters, the RKN provides an elegant way to achieve action conditioning within its recurrent cell by leveraging additive interactions between the current latent state and the action variables. We present two architectures, one for forward model learning and one for inverse model learning. Both architectures significantly outperform existing model learning frameworks as well as analytical models in terms of prediction performance on a variety of real robot dynamics models.

Cite this Paper

BibTeX
@InProceedings{pmlr-v155-shaj21a, title = {Action-Conditional Recurrent Kalman Networks For Forward and Inverse Dynamics Learning}, author = {Shaj, Vaisakh and Becker, Philipp and B\"{u}chler, Dieter and Pandya, Harit and Duijkeren, Niels van and Taylor, C. James and Hanheide, Marc and Neumann, Gerhard}, booktitle = {Proceedings of the 2020 Conference on Robot Learning}, pages = {765--781}, year = {2021}, editor = {Kober, Jens and Ramos, Fabio and Tomlin, Claire}, volume = {155}, series = {Proceedings of Machine Learning Research}, month = {16--18 Nov}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v155/shaj21a/shaj21a.pdf}, url = {https://proceedings.mlr.press/v155/shaj21a.html}, abstract = {Estimating accurate forward and inverse dynamics models is a crucial component of model-based control for sophisticated robots such as robots driven by hydraulics, artificial muscles, or robots dealing with different contact situations. Analytic models to such processes are often unavailable or inaccurate due to complex hysteresis effects, unmodelled friction and stiction phenomena, and unknown effects during contact situations. A promising approach is to obtain spatio-temporal models in a data-driven way using recurrent neural networks, as they can overcome those issues. However, such models often do not meet accuracy demands sufficiently, degenerate in performance for the required high sampling frequencies and cannot provide uncertainty estimates. We adopt a recent probabilistic recurrent neural network architecture, called Recurrent Kalman Networks (RKNs), to model learning by conditioning its transition dynamics on the control actions. RKNs outperform standard recurrent networks such as LSTMs on many state estimation tasks.Inspired by Kalman filters, the RKN provides an elegant way to achieve action conditioning within its recurrent cell by leveraging additive interactions between the current latent state and the action variables. We present two architectures, one for forward model learning and one for inverse model learning. Both architectures significantly outperform existing model learning frameworks as well as analytical models in terms of prediction performance on a variety of real robot dynamics models.} }
Endnote
%0 Conference Paper %T Action-Conditional Recurrent Kalman Networks For Forward and Inverse Dynamics Learning %A Vaisakh Shaj %A Philipp Becker %A Dieter Büchler %A Harit Pandya %A Niels van Duijkeren %A C. James Taylor %A Marc Hanheide %A Gerhard Neumann %B Proceedings of the 2020 Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2021 %E Jens Kober %E Fabio Ramos %E Claire Tomlin %F pmlr-v155-shaj21a %I PMLR %P 765--781 %U https://proceedings.mlr.press/v155/shaj21a.html %V 155 %X Estimating accurate forward and inverse dynamics models is a crucial component of model-based control for sophisticated robots such as robots driven by hydraulics, artificial muscles, or robots dealing with different contact situations. Analytic models to such processes are often unavailable or inaccurate due to complex hysteresis effects, unmodelled friction and stiction phenomena, and unknown effects during contact situations. A promising approach is to obtain spatio-temporal models in a data-driven way using recurrent neural networks, as they can overcome those issues. However, such models often do not meet accuracy demands sufficiently, degenerate in performance for the required high sampling frequencies and cannot provide uncertainty estimates. We adopt a recent probabilistic recurrent neural network architecture, called Recurrent Kalman Networks (RKNs), to model learning by conditioning its transition dynamics on the control actions. RKNs outperform standard recurrent networks such as LSTMs on many state estimation tasks.Inspired by Kalman filters, the RKN provides an elegant way to achieve action conditioning within its recurrent cell by leveraging additive interactions between the current latent state and the action variables. We present two architectures, one for forward model learning and one for inverse model learning. Both architectures significantly outperform existing model learning frameworks as well as analytical models in terms of prediction performance on a variety of real robot dynamics models.
APA
Shaj, V., Becker, P., Büchler, D., Pandya, H., Duijkeren, N.v., Taylor, C.J., Hanheide, M. & Neumann, G.. (2021). Action-Conditional Recurrent Kalman Networks For Forward and Inverse Dynamics Learning. Proceedings of the 2020 Conference on Robot Learning, in Proceedings of Machine Learning Research 155:765-781 Available from https://proceedings.mlr.press/v155/shaj21a.html.

Related Material