mlsanityData to 7 Oct 2026

DemoGen: Synthetic Demonstration Generation for Data-Efficient Visuomotor Policy Learning

arXiv 2502.16932 ↗First posted 24 Feb 2025cs.ROcited by 64

Not on the map. Why?

Abstract (arXiv)

Visuomotor policies have shown great promise in robotic manipulation but often require substantial amounts of human-collected data for effective performance. A key reason underlying the data demands is their limited spatial generalization capability, which necessitates extensive data collection across different object configurations. In this work, we present DemoGen, a low-cost, fully synthetic approach for automatic demonstration generation. Using only one human-collected demonstration per task, DemoGen generates spatially augmented demonstrations by adapting the demonstrated action trajectory to novel object configurations. Visual observations are synthesized by leveraging 3D point clouds as the modality and rearranging the subjects in the scene via 3D editing. Empirically, DemoGen significantly enhances policy performance across a diverse range of real-world manipulation tasks, showing its applicability even in challenging scenarios involving deformable objects, dexterous hand end-effectors, and bimanual platforms. Furthermore, DemoGen can be extended to enable additional out-of-distribution capabilities, including disturbance resistance and obstacle avoidance.

Key facts. Cited by 64 papers in the map (data through 7 Oct 2026). First posted on arXiv 24 Feb 2025.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 29; 2026: 35.

64 papers.

Sort:

By area:

  1. 2025GraspVLA: a Grasping Foundation Model Pre-trained on Billion-scale Synthetic Action Datacited by 120
  2. 2025DeepVerse: 4D Autoregressive Video Generation as a World Modelcited by 36
  3. 2025Parallels Between VLA Model Post-Training and Human Motor Learning: Progress, Challenges, and Trendscited by 17
  4. 2025StereoVLA: Enhancing Vision-Language-Action Models with Stereo Visioncited by 16
  5. 2025H3DP: Triply-Hierarchical Diffusion Policy for Visuomotor Learningcited by 13
  6. 2025HumanoidGen: Data Generation for Bimanual Dexterous Manipulation via LLM Reasoningcited by 12
  7. 2025VLA-Reasoner: Empowering Vision-Language-Action Models with Reasoning via Online Monte Carlo Tree Searchcited by 12
  8. 2025VGGT-DP: Generalizable Robot Control via Vision Foundation Modelscited by 11
  9. 2025Guiding Data Collection via Factored Scaling Curvescited by 10
  10. 2025DemoHLM: From One Demonstration to Generalizable Humanoid Loco-Manipulationcited by 10
  11. 2025AnchorDream: Repurposing Video Diffusion for Embodiment-Aware Robot Data Synthesiscited by 10
  12. 2026Demystifying Action Space Design for Robotic Manipulation Policiescited by 9
38 more
  1. 2026Dexterous Manipulation Policies from RGB Human Videos via 3D Hand-Object Trajectory Reconstructioncited by 8
  2. 2025MoMaGen: Generating Demonstrations under Soft and Hard Constraints for Multi-Step Bimanual Mobile Manipulationcited by 7
  3. 2025Constraint-Preserving Data Generation for Visuomotor Policy Learningcited by 6
  4. 2026Vision-Language-Action in Robotics: A Survey of Datasets, Benchmarks, and Data Enginescited by 6
  5. 2025Learning Primitive Embodied World Models: Towards Scalable Robotic Learningcited by 5
  6. 2025RESample: A Robust Data Augmentation Framework via Exploratory Sampling for Robotic Manipulationcited by 5
  7. 2026TwinRL: Digital Twin-Driven Reinforcement Learning for Real-World Robotic Manipulationcited by 5
  8. 2025R2RGEN: Real-to-Real 3D Data Generation for Spatially Generalized Manipulationcited by 4
  9. 2025FieldGen: From Teleoperated Pre-Manipulation Trajectories to Field-Guided Data Generationcited by 4
  10. 2025Real2Edit2Real: Generating Robotic Demonstrations via a 3D Control Interfacecited by 4
  11. 2025LodeStar: Long-horizon Dexterity via Synthetic Data Augmentation from Human Demonstrationscited by 3
  12. 2025UMIGen: A Unified Framework for Egocentric Point Cloud Generation and Cross-Embodiment Robotic Imitation Learningcited by 3
  13. 2025ReinforceGen: Hybrid Skill Policies with Automated Data Generation and Reinforcement Learningcited by 3
  14. 2025Learning in ImaginationLand: Omnidirectional Policies through 3D Generative Models (OP-Gen)cited by 2
  15. 2025Inference-stage Adaptation-projection Strategy Adapts Diffusion Policy to Cross-manipulators Scenarioscited by 2
  16. 2025Flow with the Force Field: Learning 3D Compliant Flow Matching Policies from Force and Demonstration-Guided Simulation Datacited by 2
  17. 2025Generalizable Hierarchical Skill Learning via Object-Centric Representationcited by 2
  18. 2025AOMGen: Photoreal, Physics-Consistent Demonstration Generation for Articulated Object Manipulationcited by 2
  19. 2026X-Distill: Cross-Architecture Vision Distillation for Visuomotor Learningcited by 2
  20. 2026FAR-Dex: Few-shot Data Augmentation and Adaptive Residual Policy Refinement for Dexterous Manipulationcited by 2
  21. 2026AffordGen: Generating Diverse Demonstrations for Generalizable Object Manipulation with Afford Correspondencecited by 2
  22. 2026WorldSample: Closed-loop Real-robot RL with World Modellingcited by 2
  23. 2026DynamicManip: Enabling Dynamic Manipulation from a Single Static Demonstrationcited by 2
  24. 2026EmbodiedSWE: Coding Agents for Long Horizon Dexterous Roboticscited by 2
  25. 2025Blindfolded Experts Generalize Better: Insights from Robotic Manipulation and Videogamescited by 1
  26. 2025IGen: Scalable Data Generation for Robot Learning from Open-World Imagescited by 1
  27. 2026ExpertGen: Scalable Sim-to-Real Expert Policy Learning from Imperfect Behavior Priorscited by 1
  28. 2026RoboDream: Compositional World Models for Scalable Robot Data Synthesiscited by 1
  29. 2026DREAM: Deployment-Time Demonstration Generation via Real-to-Sim for Scalable Policy Adaptationcited by 1
  30. 2026DATAFARM: Distribution-Aligned Task and Motion Planning for Fine-Tuning Vision-Language-Action Modelscited by 1
  31. 2025BLAZER: Bootstrapping LLM-based Manipulation Agents with Zero-Shot Data Generationcited by 0
  32. 2026Easy-IIL: Reducing Human Operational Burden in Interactive Imitation Learning via Assistant Expertscited by 0
  33. 2026Referring-Aware Visuomotor Policy Learning for Closed-Loop Manipulationcited by 0
  34. 2026GenPHRI: Agentic Generative Simulation for Physical Human-Robot Interactioncited by 0
  35. 2026ComSim: Building Scalable Real-World Robot Data Generation via Compositional Simulationcited by 0
  36. 20263D Generation for Embodied AI and Robotic Simulation: A Surveycited by 0
  37. 2026A Principled Approach for Creating High-fidelity Synthetic Demonstrations for Imitation Learningcited by 0
  38. 2026ManiSplat: Manipulation Trajectory Synthesis from Monocular Video via Decoupled 3D Gaussian Splattingcited by 0

50 most cited of 64 All 64 (JSON)