Structured Mechanical Models for Robot Learning and Control

Jayesh K. Gupta, Kunal Menda, Zachary Manchester, Mykel Kochenderfer
Proceedings of the 2nd Conference on Learning for Dynamics and Control, PMLR 120:328-337, 2020.

Abstract

Model-based methods are the dominant paradigm for controlling robotic systems, though their efficacy depends heavily on the accuracy of the model used. Deep neural networks have been used to learn models of robot dynamics from data, but they suffer from data-inefficiency and the difficulty to incorporate prior knowledge. We introduce Structured Mechanical Models, a flexible model class for mechanical systems that are data-efficient, easily amenable to prior knowledge, and easily usable with model-based control techniques. The goal of this work is to demonstrate the benefits of using Structured Mechanical Models in lieu of black-box neural networks when modeling robot dynamics. We demonstrate that they generalize better from limited data and yield more reliable model-based controllers on a variety of simulated robotic domains.

Cite this Paper

BibTeX
@InProceedings{pmlr-v120-gupta20a, title = {Structured Mechanical Models for Robot Learning and Control}, author = {Gupta, Jayesh K. and Menda, Kunal and Manchester, Zachary and Kochenderfer, Mykel}, booktitle = {Proceedings of the 2nd Conference on Learning for Dynamics and Control}, pages = {328--337}, 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/gupta20a/gupta20a.pdf}, url = {https://proceedings.mlr.press/v120/gupta20a.html}, abstract = {Model-based methods are the dominant paradigm for controlling robotic systems, though their efficacy depends heavily on the accuracy of the model used. Deep neural networks have been used to learn models of robot dynamics from data, but they suffer from data-inefficiency and the difficulty to incorporate prior knowledge. We introduce Structured Mechanical Models, a flexible model class for mechanical systems that are data-efficient, easily amenable to prior knowledge, and easily usable with model-based control techniques. The goal of this work is to demonstrate the benefits of using Structured Mechanical Models in lieu of black-box neural networks when modeling robot dynamics. We demonstrate that they generalize better from limited data and yield more reliable model-based controllers on a variety of simulated robotic domains.} }
Endnote
%0 Conference Paper %T Structured Mechanical Models for Robot Learning and Control %A Jayesh K. Gupta %A Kunal Menda %A Zachary Manchester %A Mykel Kochenderfer %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-gupta20a %I PMLR %P 328--337 %U https://proceedings.mlr.press/v120/gupta20a.html %V 120 %X Model-based methods are the dominant paradigm for controlling robotic systems, though their efficacy depends heavily on the accuracy of the model used. Deep neural networks have been used to learn models of robot dynamics from data, but they suffer from data-inefficiency and the difficulty to incorporate prior knowledge. We introduce Structured Mechanical Models, a flexible model class for mechanical systems that are data-efficient, easily amenable to prior knowledge, and easily usable with model-based control techniques. The goal of this work is to demonstrate the benefits of using Structured Mechanical Models in lieu of black-box neural networks when modeling robot dynamics. We demonstrate that they generalize better from limited data and yield more reliable model-based controllers on a variety of simulated robotic domains.
APA
Gupta, J.K., Menda, K., Manchester, Z. & Kochenderfer, M.. (2020). Structured Mechanical Models for Robot Learning and Control. Proceedings of the 2nd Conference on Learning for Dynamics and Control, in Proceedings of Machine Learning Research 120:328-337 Available from https://proceedings.mlr.press/v120/gupta20a.html.

Related Material