Identifying Mechanical Models of Unknown Objects with Differentiable Physics Simulations

Changkyu Song, Abdeslam Boularias
Proceedings of the 2nd Conference on Learning for Dynamics and Control, PMLR 120:749-760, 2020.

Abstract

This paper proposes a new method for manipulating unknown objects through a sequence of non-prehensile actions that displace an object from its initial configuration to a given goal configuration on a flat surface. The proposed method leverages recent progress in differentiable physics models to identify unknown mechanical properties of manipulated objects, such as inertia matrix, friction coefficients and external forces acting on the object. To this end, a recently proposed differentiable physics engine for two-dimensional objects is adopted in this work and extended to deal forces in the three-dimensional space. The proposed model identification technique analytically computes the gradient of the distance between forecasted poses of objects and their actual observed poses, and utilizes that gradient to search for values of the mechanical properties that reduce the reality gap.

Cite this Paper

BibTeX
@InProceedings{pmlr-v120-song20a, title = {Identifying Mechanical Models of Unknown Objects with Differentiable Physics Simulations}, author = {Song, Changkyu and Boularias, Abdeslam}, booktitle = {Proceedings of the 2nd Conference on Learning for Dynamics and Control}, pages = {749--760}, 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/song20a/song20a.pdf}, url = {https://proceedings.mlr.press/v120/song20a.html}, abstract = {This paper proposes a new method for manipulating unknown objects through a sequence of non-prehensile actions that displace an object from its initial configuration to a given goal configuration on a flat surface. The proposed method leverages recent progress in differentiable physics models to identify unknown mechanical properties of manipulated objects, such as inertia matrix, friction coefficients and external forces acting on the object. To this end, a recently proposed differentiable physics engine for two-dimensional objects is adopted in this work and extended to deal forces in the three-dimensional space. The proposed model identification technique analytically computes the gradient of the distance between forecasted poses of objects and their actual observed poses, and utilizes that gradient to search for values of the mechanical properties that reduce the reality gap.} }
Endnote
%0 Conference Paper %T Identifying Mechanical Models of Unknown Objects with Differentiable Physics Simulations %A Changkyu Song %A Abdeslam Boularias %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-song20a %I PMLR %P 749--760 %U https://proceedings.mlr.press/v120/song20a.html %V 120 %X This paper proposes a new method for manipulating unknown objects through a sequence of non-prehensile actions that displace an object from its initial configuration to a given goal configuration on a flat surface. The proposed method leverages recent progress in differentiable physics models to identify unknown mechanical properties of manipulated objects, such as inertia matrix, friction coefficients and external forces acting on the object. To this end, a recently proposed differentiable physics engine for two-dimensional objects is adopted in this work and extended to deal forces in the three-dimensional space. The proposed model identification technique analytically computes the gradient of the distance between forecasted poses of objects and their actual observed poses, and utilizes that gradient to search for values of the mechanical properties that reduce the reality gap.
APA
Song, C. & Boularias, A.. (2020). Identifying Mechanical Models of Unknown Objects with Differentiable Physics Simulations. Proceedings of the 2nd Conference on Learning for Dynamics and Control, in Proceedings of Machine Learning Research 120:749-760 Available from https://proceedings.mlr.press/v120/song20a.html.

Related Material