N-Penetrate: Active Learning of Neural Collision Handler for Complex 3D Mesh Deformations

Qingyang Tan, Zherong Pan, Breannan Smith, Takaaki Shiratori, Dinesh Manocha
Proceedings of the 39th International Conference on Machine Learning, PMLR 162:21037-21049, 2022.

Abstract

We present a robust learning algorithm to detect and handle collisions in 3D deforming meshes. We first train a neural network to detect collisions and then use a numerical optimization algorithm to resolve penetrations guided by the network. Our learned collision handler can resolve collisions for unseen, high-dimensional meshes with thousands of vertices. To obtain stable network performance in such large and unseen spaces, we apply active learning by progressively inserting new collision data based on the network inferences. We automatically label these new data using an analytical collision detector and progressively fine-tune our detection networks. We evaluate our method for collision handling of complex, 3D meshes coming from several datasets with different shapes and topologies, including datasets corresponding to dressed and undressed human poses, cloth simulations, and human hand poses acquired using multi-view capture systems.

Cite this Paper

BibTeX
@InProceedings{pmlr-v162-tan22b, title = {N-Penetrate: Active Learning of Neural Collision Handler for Complex 3{D} Mesh Deformations}, author = {Tan, Qingyang and Pan, Zherong and Smith, Breannan and Shiratori, Takaaki and Manocha, Dinesh}, booktitle = {Proceedings of the 39th International Conference on Machine Learning}, pages = {21037--21049}, year = {2022}, editor = {Chaudhuri, Kamalika and Jegelka, Stefanie and Song, Le and Szepesvari, Csaba and Niu, Gang and Sabato, Sivan}, volume = {162}, series = {Proceedings of Machine Learning Research}, month = {17--23 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v162/tan22b/tan22b.pdf}, url = {https://proceedings.mlr.press/v162/tan22b.html}, abstract = {We present a robust learning algorithm to detect and handle collisions in 3D deforming meshes. We first train a neural network to detect collisions and then use a numerical optimization algorithm to resolve penetrations guided by the network. Our learned collision handler can resolve collisions for unseen, high-dimensional meshes with thousands of vertices. To obtain stable network performance in such large and unseen spaces, we apply active learning by progressively inserting new collision data based on the network inferences. We automatically label these new data using an analytical collision detector and progressively fine-tune our detection networks. We evaluate our method for collision handling of complex, 3D meshes coming from several datasets with different shapes and topologies, including datasets corresponding to dressed and undressed human poses, cloth simulations, and human hand poses acquired using multi-view capture systems.} }
Endnote
%0 Conference Paper %T N-Penetrate: Active Learning of Neural Collision Handler for Complex 3D Mesh Deformations %A Qingyang Tan %A Zherong Pan %A Breannan Smith %A Takaaki Shiratori %A Dinesh Manocha %B Proceedings of the 39th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2022 %E Kamalika Chaudhuri %E Stefanie Jegelka %E Le Song %E Csaba Szepesvari %E Gang Niu %E Sivan Sabato %F pmlr-v162-tan22b %I PMLR %P 21037--21049 %U https://proceedings.mlr.press/v162/tan22b.html %V 162 %X We present a robust learning algorithm to detect and handle collisions in 3D deforming meshes. We first train a neural network to detect collisions and then use a numerical optimization algorithm to resolve penetrations guided by the network. Our learned collision handler can resolve collisions for unseen, high-dimensional meshes with thousands of vertices. To obtain stable network performance in such large and unseen spaces, we apply active learning by progressively inserting new collision data based on the network inferences. We automatically label these new data using an analytical collision detector and progressively fine-tune our detection networks. We evaluate our method for collision handling of complex, 3D meshes coming from several datasets with different shapes and topologies, including datasets corresponding to dressed and undressed human poses, cloth simulations, and human hand poses acquired using multi-view capture systems.
APA
Tan, Q., Pan, Z., Smith, B., Shiratori, T. & Manocha, D.. (2022). N-Penetrate: Active Learning of Neural Collision Handler for Complex 3D Mesh Deformations. Proceedings of the 39th International Conference on Machine Learning, in Proceedings of Machine Learning Research 162:21037-21049 Available from https://proceedings.mlr.press/v162/tan22b.html.

Related Material