mlsanityData to 14 Oct 2026

DART: Noise Injection for Robust Imitation Learning

arXiv 1703.09327 ↗First posted 27 Mar 2017cs.LGcited by 191

Not on the map. Why?

Cited by 118 in Offline RL and imitation learning from data, but placed by its own abstract, which didn’t fit.

Abstract (arXiv)

One approach to Imitation Learning is Behavior Cloning, in which a robot observes a supervisor and infers a control policy. A known problem with this "off-policy" approach is that the robot's errors compound when drifting away from the supervisor's demonstrations. On-policy, techniques alleviate this by iteratively collecting corrective actions for the current robot policy. However, these techniques can be tedious for human supervisors, add significant computation burden, and may visit dangerous states during training. We propose an off-policy approach that injects noise into the supervisor's policy while demonstrating. This forces the supervisor to demonstrate how to recover from errors. We propose a new algorithm, DART (Disturbances for Augmenting Robot Trajectories), that collects demonstrations with injected noise, and optimizes the noise level to approximate the error of the robot's trained policy during data collection. We compare DART with DAgger and Behavior Cloning in two domains: in simulation with an algorithmic supervisor on the MuJoCo tasks (Walker, Humanoid, Hopper, Half-Cheetah) and in physical experiments with human supervisors training a Toyota HSR robot to perform grasping in clutter. For high dimensional tasks like Humanoid, DART can be up to 3x faster in computation time and only decreases the supervisor's cumulative reward by 5% during training, whereas DAgger executes policies that have 80% less cumulative reward than the supervisor. On the grasping in clutter task, DART obtains on average a 62% performance increase over Behavior Cloning.

Key facts. Cited by 191 papers in the map (data through 14 Oct 2026). First posted on arXiv 27 Mar 2017.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2018: 2; 2019: 5; 2020: 10; 2021: 24; 2022: 25; 2023: 20; 2024: 26; 2025: 36; 2026: 43.

191 papers.

Sort:

By area:

  1. 2023Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardwarecited by 1,568
  2. 2024Diffusion Forcing: Next-token Prediction Meets Full-Sequence Diffusioncited by 681
  3. 2022BC-Z: Zero-Shot Task Generalization with Robotic Imitation Learningcited by 523
  4. 2025RoboTwin 2.0: A Scalable Data Generator and Benchmark with Strong Domain Randomization for Robust Bimanual Robotic Manipulationcited by 518
  5. 2025π*0.6: a VLA That Learns From Experiencecited by 303
  6. 2019Causal Confusion in Imitation Learningcited by 288
  7. 2019Imitation Learning as f-Divergence Minimizationcited by 123
  8. 2021Urban Driver: Learning to Drive from Real-world Demonstrations Using Policy Gradientscited by 122
  9. 2021Learning to drive from a world on railscited by 114
  10. 2020Human-in-the-Loop Imitation Learning using Remote Teleoperationcited by 112
  11. 2021ThriftyDAgger: Budget-Aware Novelty and Risk Gating for Interactive Imitation Learningcited by 108
  12. 2021S4RL: Surprisingly Simple Self-Supervision for Offline Reinforcement Learningcited by 85
38 more
  1. 2023Data Quality in Imitation Learningcited by 75
  2. 2020Safe Imitation Learning via Fast Bayesian Reward Inference from Preferencescited by 73
  3. 2020Fighting Copycat Agents in Behavioral Cloning from Observation Historiescited by 71
  4. 2022Symphony: Learning Realistic and Diverse Agents for Autonomous Driving Simulationcited by 61
  5. 2023Behavior Retrieval: Few-Shot Imitation Learning by Querying Unlabeled Datasetscited by 58
  6. 2023NeRF in the Palm of Your Hand: Corrective Augmentation for Robotics via Novel-View Synthesiscited by 57
  7. 2022Motion Policy Networkscited by 55
  8. 2023HYDRA: Hybrid Robot Actions for Imitation Learningcited by 55
  9. 2024Bidirectional Decoding: Improving Action Chunking via Guided Test-Time Samplingcited by 52
  10. 2021LazyDAgger: Reducing Context Switching in Interactive Imitation Learningcited by 51
  11. 2024DiffuseLoco: Real-Time Legged Locomotion Control with Diffusion from Offline Datasetscited by 49
  12. 2024Re-Mix: Optimizing Data Mixtures for Large Scale Imitation Learningcited by 47
  13. 2020Overcoming Model Bias for Robust Offline Deep Reinforcement Learningcited by 43
  14. 2020Grasping with Chopsticks: Combating Covariate Shift in Model-free Imitation Learning for Fine Manipulationcited by 38
  15. 2021On the Sample Complexity of Stability Constrained Imitation Learningcited by 38
  16. 2024JUICER: Data-Efficient Imitation Learning for Robotic Assemblycited by 38
  17. 2024Efficient Data Collection for Robotic Manipulation via Compositional Generalizationcited by 37
  18. 2025Much Ado About Noising: Dispelling the Myths of Generative Robotic Controlcited by 36
  19. 2024Diffusion Meets DAgger: Supercharging Eye-in-hand Imitation Learningcited by 35
  20. 2024Deep Generative Models in Robotics: A Survey on Learning from Multimodal Demonstrationscited by 35
  21. 2023Provable Guarantees for Generative Behavior Cloning: Bridging Low-Level Stability and High-Level Behaviorcited by 33
  22. 2020Imitation Learning with Sinkhorn Distancescited by 30
  23. 2022Fleet-DAgger: Interactive Robot Fleet Learning with Scalable Human Supervisioncited by 30
  24. 2023RLIF: Interactive Imitation Learning as Reinforcement Learningcited by 30
  25. 2025Taming Teacher Forcing for Masked Autoregressive Video Generationcited by 30
  26. 2023CCIL: Continuity-based Data Augmentation for Corrective Imitation Learningcited by 29
  27. 2021LILA: Language-Informed Latent Actionscited by 27
  28. 2022Robust Imitation Learning against Variations in Environment Dynamicscited by 26
  29. 2025RoboCopilot: Human-in-the-loop Interactive Imitation Learning for Robot Manipulationcited by 26
  30. 2021Keyframe-Focused Visual Imitation Learningcited by 25
  31. 2022MoCapAct: A Multi-Task Dataset for Simulated Humanoid Controlcited by 25
  32. 2024IntervenGen: Interventional Data Generation for Robust and Data-Efficient Robot Imitation Learningcited by 25
  33. 2021Learning Multi-Stage Tasks with One Demonstration via Self-Replaycited by 21
  34. 2022TaSIL: Taylor Series Imitation Learningcited by 21
  35. 2026SOP: A Scalable Online Post-Training System for Vision-Language-Action Modelscited by 21
  36. 2022Planning for Sample Efficient Imitation Learningcited by 20
  37. 2026When to Trust Imagination: Adaptive Action Execution for World Action Modelscited by 20
  38. 2022Learning Deformable Object Manipulation from Expert Demonstrationscited by 19

50 most cited of 191 All 191 (JSON)