PhysDiff: Physics-Guided Human Motion Diffusion Model

被引:17
|
作者
Yuan, Ye [1 ]
Song, Jiaming [1 ]
Iqbal, Umar [1 ]
Vahdat, Arash [1 ]
Kautz, Jan [1 ]
机构
[1] NVIDIA, Santa Clara, CA 95051 USA
关键词
D O I
10.1109/ICCV51070.2023.01467
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Denoising diffusion models hold great promise for generating diverse and realistic human motions. However, existing motion diffusion models largely disregard the laws of physics in the diffusion process and often generate physically-implausible motions with pronounced artifacts such as floating, foot sliding, and ground penetration. This seriously impacts the quality of generated motions and limits their real-world application. To address this issue, we present a novel physics-guided motion diffusion model (PhysDiff), which incorporates physical constraints into the diffusion process. Specifically, we propose a physics-based motion projection module that uses motion imitation in a physics simulator to project the denoised motion of a diffusion step to a physically-plausible motion. The projected motion is further used in the next diffusion step to guide the denoising diffusion process. Intuitively, the use of physics in our model iteratively pulls the motion toward a physically-plausible space, which cannot be achieved by simple post-processing. Experiments on large-scale human motion datasets show that our approach achieves state-of-the-art motion quality and improves physical plausibility drastically (>78% for all datasets).
引用
收藏
页码:15964 / 15975
页数:12
相关论文
共 50 条
  • [1] Physics-Guided Human Motion Capture with Pose Probability Modeling
    Ju, Jingyi
    Huang, Buzhen
    Zhu, Chen
    Li, Zhihao
    Wang, Yangang
    [J]. PROCEEDINGS OF THE THIRTY-SECOND INTERNATIONAL JOINT CONFERENCE ON ARTIFICIAL INTELLIGENCE, IJCAI 2023, 2023, : 947 - 955
  • [2] Generative Downscaling of PDE Solvers with Physics-Guided Diffusion Models
    Lu, Yulong
    Xu, Wuzhe
    [J]. Journal of Scientific Computing, 2024, 101 (03)
  • [3] A physics-guided autoregressive model for saturation sequence prediction
    Wang, Yanzhong
    Zhang, Kai
    Ma, Xiaopeng
    Liu, Piyang
    Wang, Haochen
    Guo, Xin
    Liu, Chenglong
    Zhang, Liming
    Yao, Jun
    [J]. GEOENERGY SCIENCE AND ENGINEERING, 2023, 221
  • [4] Physics-guided LSTM model for heat load prediction of buildings
    Wang, Yongjie
    Zhan, Changhong
    Li, Guanghao
    Zhang, Dongjie
    Han, Xueying
    [J]. ENERGY AND BUILDINGS, 2023, 294
  • [5] Physics-Guided Deep Generative Model For New Ligand Discovery
    Sagar, Dikshant
    Risheh, Ali
    Sheikh, Nida
    Forouzesh, Negin
    [J]. 14TH ACM CONFERENCE ON BIOINFORMATICS, COMPUTATIONAL BIOLOGY, AND HEALTH INFORMATICS, BCB 2023, 2023,
  • [6] Fusing Physics to Fiber Nonlinearity Model for Optical Networks Based on Physics-Guided Neural Networks
    Liu, Xiaomin
    Fan, Yunyun
    Zhang, Yihao
    Cai, Meng
    Liu, Lei
    Yi, Lilin
    Hu, Weisheng
    Zhuge, Qunbi
    [J]. JOURNAL OF LIGHTWAVE TECHNOLOGY, 2022, 40 (17) : 5793 - 5802
  • [7] Physics-Guided Sound Synthesis for Rotating Blades
    Xu, Siqi
    Liu, Shiguang
    [J]. ADVANCES IN COMPUTER GRAPHICS, CGI 2020, 2020, 12221 : 233 - 244
  • [8] Physics-Guided Descriptors for Prediction of Structural Polymorphs
    Grosso, Bastien F.
    Spaldin, Nicola A.
    Tehrani, Aria Mansouri
    [J]. JOURNAL OF PHYSICAL CHEMISTRY LETTERS, 2022, 13 (31): : 7342 - 7349
  • [9] Surrogate Modeling using Physics-guided Learning
    Ozdagli, Ali
    Volgyesi, Peter
    Koutsoukos, Xenofon
    [J]. 2023 CYBER-PHYSICAL SYSTEMS AND INTERNET-OF-THINGS WEEK, CPS-IOT WEEK WORKSHOPS, 2023, : 130 - 135
  • [10] A Physics-Guided Machine Learning Model for Predicting Viscoelasticity of Solids at Large Deformation
    Qin, Bao
    Zhong, Zheng
    [J]. Polymers, 2024, 16 (22)