Encoding Physical Constraints in Differentiable Newton-Euler Algorithm

Giovanni Sutanto, Austin Wang, Yixin Lin, Mustafa Mukadam, Gaurav Sukhatme, Akshara Rai, Franziska Meier
Proceedings of the 2nd Conference on Learning for Dynamics and Control, PMLR 120:804-813, 2020.

Abstract

The recursive Newton-Euler Algorithm (RNEA) is a popular technique in robotics for computing the dynamics of robots. The computed dynamics can then be used for torque control with inverse dynamics, or for forward dynamics computations. RNEA can be framed as a differentiable computational graph, enabling the dynamics parameters of the robot to be learned from data. However, the dynamics parameters learned in this manner can be physically implausible. In this work, we incorporate physical constraints in the learning by adding structure to the learned parameters. This results in a framework that can learn physically plausible dynamics, improving the training speed as well as generalization of the learned dynamics models. We evaluate our method on real-time inverse dynamics predictions of a 7 degree of freedom robot arm, both in simulation and on the real robot. Our experiments study a spectrum of structure added to learned dynamics, and compare their performance and generalization.

Cite this Paper

BibTeX
@InProceedings{pmlr-v120-sutanto20a, title = {Encoding Physical Constraints in Differentiable Newton-Euler Algorithm}, author = {Sutanto, Giovanni and Wang, Austin and Lin, Yixin and Mukadam, Mustafa and Sukhatme, Gaurav and Rai, Akshara and Meier, Franziska}, booktitle = {Proceedings of the 2nd Conference on Learning for Dynamics and Control}, pages = {804--813}, year = {2020}, editor = {Bayen, Alexandre M. and Jadbabaie, Ali and Pappas, George and Parrilo, Pablo A. and Recht, Benjamin and Tomlin, Claire and Zeilinger, Melanie}, volume = {120}, series = {Proceedings of Machine Learning Research}, month = {10--11 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v120/sutanto20a/sutanto20a.pdf}, url = {https://proceedings.mlr.press/v120/sutanto20a.html}, abstract = {The recursive Newton-Euler Algorithm (RNEA) is a popular technique in robotics for computing the dynamics of robots. The computed dynamics can then be used for torque control with inverse dynamics, or for forward dynamics computations. RNEA can be framed as a differentiable computational graph, enabling the dynamics parameters of the robot to be learned from data. However, the dynamics parameters learned in this manner can be physically implausible. In this work, we incorporate physical constraints in the learning by adding structure to the learned parameters. This results in a framework that can learn physically plausible dynamics, improving the training speed as well as generalization of the learned dynamics models. We evaluate our method on real-time inverse dynamics predictions of a 7 degree of freedom robot arm, both in simulation and on the real robot. Our experiments study a spectrum of structure added to learned dynamics, and compare their performance and generalization.} }
Endnote
%0 Conference Paper %T Encoding Physical Constraints in Differentiable Newton-Euler Algorithm %A Giovanni Sutanto %A Austin Wang %A Yixin Lin %A Mustafa Mukadam %A Gaurav Sukhatme %A Akshara Rai %A Franziska Meier %B Proceedings of the 2nd Conference on Learning for Dynamics and Control %C Proceedings of Machine Learning Research %D 2020 %E Alexandre M. Bayen %E Ali Jadbabaie %E George Pappas %E Pablo A. Parrilo %E Benjamin Recht %E Claire Tomlin %E Melanie Zeilinger %F pmlr-v120-sutanto20a %I PMLR %P 804--813 %U https://proceedings.mlr.press/v120/sutanto20a.html %V 120 %X The recursive Newton-Euler Algorithm (RNEA) is a popular technique in robotics for computing the dynamics of robots. The computed dynamics can then be used for torque control with inverse dynamics, or for forward dynamics computations. RNEA can be framed as a differentiable computational graph, enabling the dynamics parameters of the robot to be learned from data. However, the dynamics parameters learned in this manner can be physically implausible. In this work, we incorporate physical constraints in the learning by adding structure to the learned parameters. This results in a framework that can learn physically plausible dynamics, improving the training speed as well as generalization of the learned dynamics models. We evaluate our method on real-time inverse dynamics predictions of a 7 degree of freedom robot arm, both in simulation and on the real robot. Our experiments study a spectrum of structure added to learned dynamics, and compare their performance and generalization.
APA
Sutanto, G., Wang, A., Lin, Y., Mukadam, M., Sukhatme, G., Rai, A. & Meier, F.. (2020). Encoding Physical Constraints in Differentiable Newton-Euler Algorithm. Proceedings of the 2nd Conference on Learning for Dynamics and Control, in Proceedings of Machine Learning Research 120:804-813 Available from https://proceedings.mlr.press/v120/sutanto20a.html.

Related Material