Collision-aware interactive simulation using graph neural networks
Tóm tắt
Deep simulations have gained widespread attention owing to their excellent acceleration performances. However, these methods cannot provide effective collision detection and response strategies. We propose a deep interactive physical simulation framework that can effectively address tool-object collisions. The framework can predict the dynamic information by considering the collision state. In particular, the graph neural network is chosen as the base model, and a collision-aware recursive regression module is introduced to update the network parameters recursively using interpenetration distances calculated from the vertex-face and edge-edge tests. Additionally, a novel self-supervised collision term is introduced to provide a more compact collision response. This study extensively evaluates the proposed method and shows that it effectively reduces interpenetration artifacts while ensuring high simulation efficiency.
Tài liệu tham khảo
Holden D, Duong BC, Datta S, Nowrouzezahrai D (2019) Subspace neural physics: Fast data-driven interactive simulation. Paper presented at the 18th annual ACM SIGGRAPH/Eurographics symposium on computer animation, ACM, Los Angeles, 26-28 July 2019. https://doi.org/10.1145/3309486.3340245
Santesteban I, Thuerey N, Otaduy MA, Casas D (2021) Self-supervised collision handling via generative 3D garment models for virtual try-on. Paper presented at the 2021 IEEE/CVF conference on computer vision and pattern recognition, IEEE, Nashville, 20-25 June 2021. https://doi.org/10.1109/CVPR46437.2021.01159
Luo R, Shao TJ, Wang HM, Xu WW, Chen X, Zhou K et al (2020) NNWarp: Neural network-based nonlinear deformation. IEEE Trans Vis Comput Graph 26(4):1745-1759.
Romero C, Casas D, Pérez J, Otaduy M (2021) Learning contact corrections for handle-based subspace dynamics. ACM Trans Graph 40(4): 131. https://doi.org/10.1145/3476576.3476703
Teng Y, Meyer M, DeRose T, Kim T (2015) Subspace condensation: Full space adaptivity for subspace deformations. ACM Trans Graph 34(4): 76. https://doi.org/10.1145/2766904
Tan QY, Pan ZR, Manocha D (2021) L Collision: Fast generation of collision-free human poses using learned non-penetration constraints. arXiv preprint arXiv:2011.03632.
Pfaff T, Fortunato M, Sanchez-Gonzalez A, Battaglia PW (2021) Learning mesh-based simulation with graph networks. arXiv preprint arXiv:2010.03409.
Coming DS, Staadt OG (2008) Velocity-aligned discrete oriented polytopes for dynamic collision detection. IEEE Trans Vis Comput Graph 14(1):1-12. https://doi.org/10.1109/TVCG.2007.70405
Redon S, Kheddar A, Coquillart S (2002) Fast continuous collision detection between rigid bodies. Comput Graph Forum 21(3):279-287. https://doi.org/10.1111/1467-8659.t01-1-00587
Redon S, Lin MC (2006) A fast method for local penetration depth computation. J Graph Tools 11(2):37-50. https://doi.org/10.1080/2151237X.2006.10129216
Tang M, Manocha D, Otaduy MA, Tong RF (2012) Continuous penalty forces. ACM Trans Graph 31(4): 107. https://doi.org/10.1145/2185520.2185603
Choi YK, Wang WP, Mourrain B, Tu CH, Jia XH, Sun F (2014) Continuous collision detection for composite quadric models. Graph Models 76(5):566-579. https://doi.org/10.1016/j.gmod.2014.03.005
Battaglia PW, Hamrick JB, Bapst V, Sanchez-Gonzalez A, Zambaldi V, Malinowski M et al (2018) Relational inductive biases, deep learning, and graph networks. arXiv preprint arXiv:1806.01261.
Kipf T, Fetaya E, Wang KC, Welling M, Zemel R (2018) Neural relational inference for interacting systems. Paper presented at the 35th international conference on machine learning, JMLR.org. Stockholm; 2018.
Gilmer J, Schoenholz SS, Riley PF, Vinyals O, Dahl GE (2017) Neural message passing for quantum chemistry. Paper presented at the 34th international conference on machine learning, JMLR.org. Sydney; 2017.
Seo S, Liu Y (2019) Differentiable physics-informed graph networks. arXiv preprint arXiv:1902.02950.
Chang MB, Ullman T, Torralba A, Tenenbaum JB (2017) A compositional object-based approach to learning physical dynamics. arXiv preprint arXiv:1612.00341.
Sanchez-Gonzalez A, Heess N, Springenberg JT, Merel J, Riedmiller M, Hadsell R. et al (2018) Graph networks as learnable physics engines for inference and control. Paper presented at the 35th international conference on machine learning, JMLR.org. Stockholm; 2018.
Li YZ, Wu JJ, Zhu JY, Tenenbaum JB, Torralba A, Tedrake R (2019) Propagation networks for model-based control under partial observation. Paper presented at the 2019 international conference on robotics and automation, IEEE, Montrl, 20-24 May 2019. https://doi.org/10.1109/ICRA.2019.8793509
Mrowca D, Zhuang CX, Wang E, Haber N, Fei-Fei L, Tenenbaum JB et al (2018) Flexible neural representation for physics prediction. arXiv preprint arXiv:1806.08047.
Sanchez-Gonzalez A, Bapst V, Cranmer K, Battaglia P (2019) Hamiltonian graph networks with ODE integrators. arXiv preprint arXiv:1909.12790.
Sanchez-Gonzalez A, Godwin J, Pfaff T, Ying R, Leskovec J, Battaglia P (2020) Learning to simulate complex physics with graph networks. Paper presented at the 37th international conference on machine learning, PMLR, Vienna.
Tang M, Manocha D, Tong RF (2010) Fast continuous collision detection using deforming non-penetration filters. Paper presented at the 2010 ACM SIGGRAPH symposium on interactive 3D graphics and games, ACM, Washington. https://doi.org/10.1145/1730804.1730806
Kingma DP, Welling M (2014) Auto-encoding variational bayes. arXiv preprint arXiv:1312.6114.
Li MC, Ferguson Z, Schneider T, Langlois T, Zorin D, Panozzo D et al (2020) Incremental potential contact: Intersection-and inversion-free, large-deformation dynamics. ACM Trans Graph 39(4): 49. https://doi.org/10.1145/3386569.3392425
Kingma DB, Ba J (2017) Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980.