mlsanityData to 14 Oct 2026

Novel Demonstration Generation with Gaussian Splatting Enables Robust One-Shot Manipulation

arXiv 2504.13175 ↗First posted 17 Apr 2025cs.ROcited by 48

Not on the map. Why?

Abstract (arXiv)

Visuomotor policies learned from teleoperated demonstrations face challenges such as lengthy data collection, high costs, and limited data diversity. Existing approaches address these issues by augmenting image observations in RGB space or employing Real-to-Sim-to-Real pipelines based on physical simulators. However, the former is constrained to 2D data augmentation, while the latter suffers from imprecise physical simulation caused by inaccurate geometric reconstruction. This paper introduces RoboSplat, a novel method that generates diverse, visually realistic demonstrations by directly manipulating 3D Gaussians. Specifically, we reconstruct the scene through 3D Gaussian Splatting (3DGS), directly edit the reconstructed scene, and augment data across six types of generalization with five techniques: 3D Gaussian replacement for varying object types, scene appearance, and robot embodiments; equivariant transformations for different object poses; visual attribute editing for various lighting conditions; novel view synthesis for new camera perspectives; and 3D content generation for diverse object types. Comprehensive real-world experiments demonstrate that RoboSplat significantly enhances the generalization of visuomotor policies under diverse disturbances. Notably, while policies trained on hundreds of real-world demonstrations with additional 2D data augmentation achieve an average success rate of 57.2%, RoboSplat attains 87.8% in one-shot settings across six types of generalization in the real world.

Key facts. Cited by 48 papers in the map (data through 14 Oct 2026). First posted on arXiv 17 Apr 2025.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 21; 2026: 27.

48 papers.

Sort:

By area:

  1. 2025GraspVLA: a Grasping Foundation Model Pre-trained on Billion-scale Synthetic Action Datacited by 123
  2. 2025Real-to-Sim Robot Policy Evaluation with Gaussian Splatting Simulation of Soft-Body Interactionscited by 30
  3. 2025GSWorld: Closed-Loop Photo-Realistic Simulation Suite for Robotic Manipulationcited by 19
  4. 2025StereoVLA: Enhancing Vision-Language-Action Models with Stereo Visioncited by 18
  5. 2025Do You Know Where Your Camera Is? View-Invariant Policy Learning with Camera Conditioningcited by 14
  6. 2025ManipDreamer3D : Synthesizing Plausible Robotic Manipulation Video with Occupancy-aware 3D Trajectorycited by 12
  7. 2025EMMA: Generalizing Real-World Robot Manipulation via Generative Visual Transfercited by 12
  8. 2025VGGT-DP: Generalizable Robot Control via Vision Foundation Modelscited by 11
  9. 2025OXE-AugE: A Large-Scale Robot Augmentation of OXE for Scaling Cross-Embodiment Policy Learningcited by 10
  10. 2025TwinAligner: Visual-Dynamic Alignment Empowers Physics-aware Real2Sim2Real for Robotic Manipulationcited by 7
  11. 2026RoboVIP: Multi-View Video Generation with Visual Identity Prompting Augments Robot Manipulationcited by 7
  12. 2026Reconstruction by Generation: 3D Multi-Object Scene Reconstruction from Sparse Observationscited by 7
36 more
  1. 2025Learning Primitive Embodied World Models: Towards Scalable Robotic Learningcited by 6
  2. 2025See Once, Then Act: Vision-Language-Action Model with Task Learning from One-Shot Video Demonstrationscited by 6
  3. 2026TwinRL: Digital Twin-Driven Reinforcement Learning for Real-World Robotic Manipulationcited by 6
  4. 2026ViTaS: Visual Tactile Soft Fusion Contrastive Learning for Visuomotor Learningcited by 5
  5. 2025ViDAR: Video Diffusion-Aware 4D Reconstruction From Monocular Inputscited by 4
  6. 2025OMP: One-step Meanflow Policy with Directional Alignmentcited by 4
  7. 2025Real2Edit2Real: Generating Robotic Demonstrations via a 3D Control Interfacecited by 4
  8. 2026CRAFT: Video Diffusion for Bimanual Robot Data Generationcited by 4
  9. 2025ROPA: Synthetic Robot Pose Generation for RGB-D Bimanual Data Augmentationcited by 3
  10. 2026D-REX: Differentiable Real-to-Sim-to-Real Engine for Learning Dexterous Graspingcited by 3
  11. 2026RoboDream: Compositional World Models for Scalable Robot Data Synthesiscited by 3
  12. 2025IGFuse: Interactive 3D Gaussian Scene Reconstruction via Multi-Scans Fusioncited by 2
  13. 2025Learning in ImaginationLand: Omnidirectional Policies through 3D Generative Models (OP-Gen)cited by 2
  14. 2025Observer-Actor: Active Vision Imitation Learning with Sparse-View Gaussian Splattingcited by 2
  15. 2025AOMGen: Photoreal, Physics-Consistent Demonstration Generation for Articulated Object Manipulationcited by 2
  16. 2026M^3: Dense Matching Meets Multi-View Foundation Models for Monocular Gaussian Splatting SLAMcited by 2
  17. 2026An Efficient Metric for Data Quality Measurement in Imitation Learningcited by 2
  18. 2026DynamicManip: Enabling Dynamic Manipulation from a Single Static Demonstrationcited by 2
  19. 2026DREAM: Deployment-Time Demonstration Generation via Real-to-Sim for Scalable Policy Adaptationcited by 2
  20. 2025DSG-World: Learning a 3D Gaussian World Model from Dual State Videoscited by 1
  21. 2026Gaussian Sequences with Multi-Scale Dynamics for 4D Reconstruction from Monocular Casual Videoscited by 1
  22. 2026SIM1: Physics-Aligned Simulator as Zero-Shot Data Scaler in Deformable Worldscited by 1
  23. 20263D Generation for Embodied AI and Robotic Simulation: A Surveycited by 1
  24. 2026FrozenDrive: Zero-Shot Text-Guided Driving Scene Generation and Data Augmentation with Parameter-Free Frozen Diffusion Modelcited by 1
  25. 2026AnchorVLA4D: an Anchor-Based Spatial-Temporal Vision-Language-Action Model for Robotic Manipulationcited by 0
  26. 2026A Principled Approach for Creating High-fidelity Synthetic Demonstrations for Imitation Learningcited by 0
  27. 2026ManiSplat: Manipulation Trajectory Synthesis from Monocular Video via Decoupled 3D Gaussian Splattingcited by 0
  28. 2026UniviewVLA: A Unified Multiview Vision-Language-Action Model with World Modelingcited by 0
  29. 2026FalconTrack: Photorealistic Auto-Labeled Perception and Physics-Aware Vision-Based Aerial Trackingcited by 0
  30. 2026Domain Arithmetic: One-Shot VLA Adaptation under Environmental Shiftscited by 0
  31. 2026The Moving Eye: Enhancing VLA Spatial Generalization via Hybrid Dynamic Data Collectioncited by 0
  32. 2026InfiNoVA: Infinite Novel View Augmentation for Viewpoint Invariant Robot Policiescited by 0
  33. 2026F4R: Failure-Driven Recognition, Reconstruction, Refinement, and Redeployment for Continual Robot Self-Improvementcited by 0
  34. 2026MobileVISTA: Generative Data Augmentation for Pose Generalization in Mobile Manipulationcited by 0
  35. 2026StereoEngine: Scaling Stereo Bimanual Robot Data without Stereo Camerascited by 0
  36. 2026DexNinja: Learning Robust Dexterous Cutting Policy with a Real-to-Sim-to-Real Data Enginecited by 0