Realtime Simulation of Thin-Shell Deformable Materials Using CNN-Based Mesh Embedding

doi:10.1109/LRA.2020.2970624

CORC > 计算技术研究所 > 中国科学院计算技术研究所 > 中国科学院计算技术研究所期刊论文 > 英文

	Realtime Simulation of Thin-Shell Deformable Materials Using CNN-Based Mesh Embedding
	Tan, Qingyang 2; Pan, Zherong 1; Gao, Lin 3; Manocha, Dinesh 2
刊名	IEEE ROBOTICS AND AUTOMATION LETTERS
	2020-04-01
卷号	5 期号:2 页码:2325-2332
关键词	Simulation and animation dexterous manipulation
ISSN号	2377-3766
DOI	10.1109/LRA.2020.2970624
英文摘要	We address the problem of accelerating thin-shell deformable object simulations by dimension reduction. We present a new algorithm to embed a high-dimensional configuration space of deformable objects in a low-dimensional feature space, where the configurations of objects and feature points have approximate one-to-one mapping. Our key technique is a graph-based convolutional neural network (CNN) defined on meshes with arbitrary topologies and a new mesh embedding approach based on physics-inspired loss term. We have applied our approach to accelerate high-resolution thin shell simulations corresponding to cloth-like materials, where the configuration space has tens of thousands of degrees of freedom. We show that our physics-inspired embedding approach leads to higher accuracy compared with prior mesh embedding methods. Finally, we show that the temporal evolution of the mesh in the feature space can also be learned using a recurrent neural network (RNN) leading to fully learnable physics simulators. After training our learned simulator runs 500-10000x faster and the accuracy is high enough for robot manipulation tasks.
资助项目	ARO[W911NF1810313] ; ARO[W911NF1910315] ; Intel ; National Natural Science Foundation of China[61872440] ; Beijing Municipal Natural Science Foundation[L182016]
WOS研究方向	Robotics
语种	英语
出版者	IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
WOS记录号	WOS:000526572000030
内容类型	期刊论文
源URL	[http://119.78.100.204/handle/2XEOYT63/14209]
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Pan, Zherong
作者单位	1.Univ N Carolina, Dept Comp Sci, Chapel Hill, NC 27514 USA 2.Univ Maryland, Dept Comp Sci & Elect & Comp Engn, College Pk, MD 20740 USA 3.Chinese Acad Sci, Inst Comp Technol, Beijing, Peoples R China
推荐引用方式 GB/T 7714	Tan, Qingyang,Pan, Zherong,Gao, Lin,et al. Realtime Simulation of Thin-Shell Deformable Materials Using CNN-Based Mesh Embedding[J]. IEEE ROBOTICS AND AUTOMATION LETTERS,2020,5(2):2325-2332.
APA	Tan, Qingyang,Pan, Zherong,Gao, Lin,&Manocha, Dinesh.(2020).Realtime Simulation of Thin-Shell Deformable Materials Using CNN-Based Mesh Embedding.IEEE ROBOTICS AND AUTOMATION LETTERS,5(2),2325-2332.
MLA	Tan, Qingyang,et al."Realtime Simulation of Thin-Shell Deformable Materials Using CNN-Based Mesh Embedding".IEEE ROBOTICS AND AUTOMATION LETTERS 5.2(2020):2325-2332.