mlsanityData to 7 Oct 2026

Self-Supervised Correspondence in Visuomotor Policy Learning

arXiv 1909.06933 ↗First posted 16 Sep 2019cs.RO cs.CV cs.LGcited by 106

On the map: Offline RL and imitation learning from data → Robot manipulation imitation learning

Abstract (arXiv)

In this paper we explore using self-supervised correspondence for improving the generalization performance and sample efficiency of visuomotor policy learning. Prior work has primarily used approaches such as autoencoding, pose-based losses, and end-to-end policy optimization in order to train the visual portion of visuomotor policies. We instead propose an approach using self-supervised dense visual correspondence training, and show this enables visuomotor policy learning with surprisingly high generalization performance with modest amounts of data: using imitation learning, we demonstrate extensive hardware validation on challenging manipulation tasks with as few as 50 demonstrations. Our learned policies can generalize across classes of objects, react to deformable object configurations, and manipulate textureless symmetrical objects in a variety of backgrounds, all with closed-loop, real-time vision-based policies. Simulated imitation learning experiments suggest that correspondence training offers sample complexity and generalization benefits compared to autoencoding and end-to-end training.

Key facts. Cited by 106 papers in the map (data through 7 Oct 2026). Placed in Robot manipulation imitation learning, one of 13 directions in Offline RL and imitation learning from data. First posted on arXiv 16 Sep 2019.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2020: 2; 2021: 14; 2022: 19; 2023: 24; 2024: 20; 2025: 20; 2026: 7.

106 papers.

Sort:

By area:

  1. 2023Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardwarecited by 1,490
  2. 2021What Matters in Learning from Offline Human Demonstrations for Robot Manipulationcited by 699
  3. 2021CLIPort: What and Where Pathways for Robotic Manipulationcited by 617
  4. 2022Socratic Models: Composing Zero-Shot Multimodal Reasoning with Languagecited by 513
  5. 2021Implicit Behavioral Cloningcited by 442
  6. 2022VIMA: General Robot Manipulation with Multimodal Promptscited by 403
  7. 2022Behavior Transformers: Cloning k modes with one stonecited by 299
  8. 2022TAP-Vid: A Benchmark for Tracking Any Point in a Videocited by 258
  9. 2024Diffusion Policy Policy Optimizationcited by 255
  10. 2022Real-World Robot Learning with Masked Visual Pre-trainingcited by 241
  11. 2023MimicPlay: Long-Horizon Imitation Learning by Watching Human Playcited by 214
  12. 2021How to Train Your Robot with Deep Reinforcement Learning; Lessons We've Learnedcited by 201
38 more
  1. 2024DexCap: Scalable and Portable Mocap Data Collection System for Dexterous Manipulationcited by 149
  2. 2022VIOLA: Imitation Learning for Vision-Based Manipulation with Object Proposal Priorscited by 135
  3. 2024Reconciling Reality through Simulation: A Real-to-Sim-to-Real Approach for Robust Manipulationcited by 135
  4. 2023Distilled Feature Fields Enable Few-Shot Language-Guided Manipulationcited by 130
  5. 2025Spatial Forcing: Implicit Spatial Representation Alignment for Vision-language-action Modelcited by 125
  6. 2021XIRL: Cross-embodiment Inverse Reinforcement Learningcited by 101
  7. 2021Coarse-to-Fine Imitation Learning: Robot Manipulation from a Single Demonstrationcited by 92
  8. 2022NeRF-Supervision: Learning Dense Object Descriptors from Neural Radiance Fieldscited by 86
  9. 2021S4RL: Surprisingly Simple Self-Supervision for Offline Reinforcement Learningcited by 85
  10. 2024Dreamitate: Real-World Visuomotor Policy Learning via Video Generationcited by 84
  11. 2024General Flow as Foundation Affordance for Scalable Robot Learningcited by 80
  12. 2022You Only Demonstrate Once: Category-Level Manipulation from Single Visual Demonstrationcited by 79
  13. 2021VAT-Mart: Learning Visual Action Trajectory Proposals for Manipulating 3D ARTiculated Objectscited by 78
  14. 2023XSkill: Cross Embodiment Skill Discoverycited by 74
  15. 2024Unpacking Failure Modes of Generative Policies: Runtime Monitoring of Consistency and Progresscited by 63
  16. 2024Equivariant Diffusion Policycited by 61
  17. 2023Learning Generalizable Manipulation Policies with Object-Centric 3D Representationscited by 60
  18. 2023Waypoint-Based Imitation Learning for Robotic Manipulationcited by 59
  19. 2023NeRF in the Palm of Your Hand: Corrective Augmentation for Robotics via Novel-View Synthesiscited by 56
  20. 2022SE(3)-Equivariant Relational Rearrangement with Neural Descriptor Fieldscited by 51
  21. 2023Robot Learning with Sensorimotor Pre-trainingcited by 51
  22. 2022Graph Inverse Reinforcement Learning from Diverse Videoscited by 45
  23. 2023Crossway Diffusion: Improving Diffusion-based Visuomotor Policy via Self-supervised Learningcited by 37
  24. 2024Diffusion Meets DAgger: Supercharging Eye-in-hand Imitation Learningcited by 34
  25. 2024ARCap: Collecting High-quality Human Demonstrations for Robot Learning with Augmented Reality Feedbackcited by 31
  26. 2023Shelving, Stacking, Hanging: Relational Pose Diffusion for Multi-modal Rearrangementcited by 29
  27. 2022K-VIL: Keypoints-based Visual Imitation Learningcited by 27
  28. 2022USEEK: Unsupervised SE(3)-Equivariant 3D Keypoints for Generalizable Manipulationcited by 27
  29. 2023CCIL: Continuity-based Data Augmentation for Corrective Imitation Learningcited by 26
  30. 2021Generalization Through Hand-Eye Coordination: An Action Space for Learning Spatially-Invariant Visuomotor Controlcited by 24
  31. 2021Learning Multi-Stage Tasks with One Demonstration via Self-Replaycited by 21
  32. 2022Visuomotor Control in Multi-Object Scenes Using Object-Aware Representationscited by 21
  33. 2024Don't Start from Scratch: Behavioral Refinement via Interpolant-based Policy Diffusioncited by 21
  34. 2025SKIL: Semantic Keypoint Imitation Learning for Generalizable Data-efficient Manipulationcited by 21
  35. 20233D Implicit Transporter for Temporally Consistent Keypoint Discoverycited by 20
  36. 2023Chain-of-Thought Predictive Controlcited by 18
  37. 2024Cross-Embodiment Robot Manipulation Skill Transfer using Latent Space Alignmentcited by 17
  38. 2023End-to-End (Instance)-Image Goal Navigation through Correspondence as an Emergent Phenomenoncited by 16

50 most cited of 106 All 106 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2025Correspondence-Oriented Imitation Learning: Flexible Visuomotor Control with 3D Conditioning
  2. 2018Reinforcement and Imitation Learning for Diverse Visuomotor Skills
  3. 2025DINOv3-Diffusion Policy: Self-Supervised Large Visual Model for Visuomotor Diffusion Policy Learning
  4. 2025A Study on Enhancing the Generalization Ability of Visuomotor Policies via Data Augmentation ↗
  5. 2025SViP: Sequencing Bimanual Visuomotor Policies with Object-Centric Motion Primitives
  6. 2024DynaMo: In-Domain Dynamics Pretraining for Visuo-Motor Control
  7. 2026AdaTempo: Learning Shared Relative Tempo from Demonstrations for Faster Robot Manipulation
  8. 2023SpawnNet: Learning Generalizable Visuomotor Skills from Pre-trained Networks