mlsanityData to 7 Oct 2026

ContactNets: Learning Discontinuous Contact Dynamics with Smooth, Implicit Representations

arXiv 2009.11193 ↗First posted 23 Sep 2020cs.RO cs.LGcited by 38

Not on the map. Why?

Abstract (arXiv)

Common methods for learning robot dynamics assume motion is continuous, causing unrealistic model predictions for systems undergoing discontinuous impact and stiction behavior. In this work, we resolve this conflict with a smooth, implicit encoding of the structure inherent to contact-induced discontinuities. Our method, ContactNets, learns parameterizations of inter-body signed distance and contact-frame Jacobians, a representation that is compatible with many simulation, control, and planning environments for robotics. We furthermore circumvent the need to differentiate through stiff or non-smooth dynamics with a novel loss function inspired by the principles of complementarity and maximum dissipation. Our method can predict realistic impact, non-penetration, and stiction when trained on 60 seconds of real-world data.

Key facts. Cited by 38 papers in the map (data through 7 Oct 2026). First posted on arXiv 23 Sep 2020.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 7; 2022: 4; 2023: 6; 2024: 12; 2025: 6; 2026: 3.

38 papers.

Sort:

By area:

  1. 2021Neural Fields in Visual Computing and Beyondcited by 539
  2. 2021Implicit Behavioral Cloningcited by 442
  3. 2022Learning Multi-Object Dynamics with Compositional Neural Radiance Fieldscited by 63
  4. 2022Enforcing Hard Constraints with Soft Barriers: Safe Reinforcement Learning in Unknown Stochastic Environmentscited by 54
  5. 2024Does equivariance matter at scale?cited by 49
  6. 2022Differentiable Physics Simulation of Dynamics-Augmented Neural Objectscited by 45
  7. 2022Learning rigid dynamics with face interaction graph networkscited by 43
  8. 2021Learning Models as Functionals of Signed-Distance Fields for Manipulation Planningcited by 29
  9. 2023Robust and Versatile Bipedal Jumping Control through Reinforcement Learningcited by 27
  10. 2021KNODE-MPC: A Knowledge-based Data-driven Predictive Control Framework for Aerial Robotscited by 26
  11. 2023CCIL: Continuity-based Data Augmentation for Corrective Imitation Learningcited by 26
  12. 2024A Survey of Optimization-based Task and Motion Planning: From Classical To Learning Approachescited by 19
26 more
  1. 2023Chain-of-Thought Predictive Controlcited by 18
  2. 2025ENFORCE: Nonlinear Constrained Learning with Adaptive-depth Neural Projectioncited by 15
  3. 2024Learning to Walk from Three Minutes of Real-World Data with Semi-structured Dynamics Modelscited by 14
  4. 2023Partition Speeds Up Learning Implicit Neural Representations Based on Exponential-Increase Hypothesiscited by 13
  5. 2024Building Minimal and Reusable Causal State Abstractions for Reinforcement Learningcited by 11
  6. 2024QuasiSim: Parameterized Quasi-Physical Simulators for Dexterous Manipulations Transfercited by 10
  7. 2025From Screens to Scenes: A Survey of Embodied AI in Healthcarecited by 10
  8. 2025Explorable INR: An Implicit Neural Representation for Ensemble Simulation Enabling Efficient Spatial and Parameter Explorationcited by 8
  9. 2021OSCAR: Data-Driven Operational Space Control for Adaptive and Robust Robot Manipulationcited by 7
  10. 2023Differentiable Visual Computing for Inverse Problems and Machine Learningcited by 7
  11. 2024Verified Safe Reinforcement Learning for Neural Network Dynamic Modelscited by 7
  12. 2025Neural Robot Dynamicscited by 7
  13. 2026Manifold-Constrained Energy-Based Transition Models for Offline Reinforcement Learningcited by 6
  14. 2024LeTO: Learning Constrained Visuomotor Policy with Differentiable Trajectory Optimizationcited by 5
  15. 2024One-Shot Real-to-Sim via End-to-End Differentiable Simulation and Renderingcited by 5
  16. 2025Differentiable Simulation of Hard Contacts with Soft Gradients for Learning and Controlcited by 5
  17. 2024TSINR: Capturing Temporal Continuity via Implicit Neural Representations for Time Series Anomaly Detectioncited by 4
  18. 2021Learning Linear Complementarity Systemscited by 3
  19. 2024Visual Imitation Learning of Task-Oriented Object Grasping and Rearrangementcited by 3
  20. 2026Simulation Distillation: Pretraining World Models in Simulation for Rapid Real-World Adaptationcited by 3
  21. 2024Improving Efficiency of Iso-Surface Extraction on Implicit Neural Representations Using Uncertainty Propagationcited by 2
  22. 2021Learning Reactive and Predictive Differentiable Controllers for Switching Linear Dynamical Modelscited by 1
  23. 2024MBDS: A Multi-Body Dynamics Simulation Dataset for Graph Networks Simulatorscited by 1
  24. 2025CHyLL: Learning Continuous Neural Representations of Hybrid Systemscited by 1
  25. 2023Instance-Agnostic Geometry and Contact Dynamics Learningcited by 0
  26. 2026Amortized Low-Rank Adaptation for Model-Based Reinforcement Learningcited by 0