mlsanityData to 7 Oct 2026

Keyframe-Focused Visual Imitation Learning

arXiv 2106.06452 ↗First posted 11 Jun 2021cs.LG cs.ROcited by 25

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

Abstract (arXiv)

Imitation learning trains control policies by mimicking pre-recorded expert demonstrations. In partially observable settings, imitation policies must rely on observation histories, but many seemingly paradoxical results show better performance for policies that only access the most recent observation. Recent solutions ranging from causal graph learning to deep information bottlenecks have shown promising results, but failed to scale to realistic settings such as visual imitation. We propose a solution that outperforms these prior approaches by upweighting demonstration keyframes corresponding to expert action changepoints. This simple approach easily scales to complex visual imitation settings. Our experimental results demonstrate consistent performance improvements over all baselines on image-based Gym MuJoCo continuous control tasks. Finally, on the CARLA photorealistic vision-based urban driving simulator, we resolve a long-standing issue in behavioral cloning for driving by demonstrating effective imitation from observation histories. Supplementary materials and code at: urlhttps://tinyurl.com/imitation-keyframes.

Key facts. Cited by 25 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. Builds on 24 of its 42 references that are in the map. First posted on arXiv 11 Jun 2021.

Builds on

24 of 42 references.

References · 24

  1. 2017Proximal Policy Optimization Algorithmscited by 14,441
  2. 2015Trust Region Policy Optimizationcited by 2,503
  3. 2017CARLA: An Open Urban Driving Simulatorcited by 2,431
  4. 2020Shortcut Learning in Deep Neural Networkscited by 1,928
  5. 2016Generative Adversarial Imitation Learningcited by 1,576
  6. 2018ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustnesscited by 1,430
  7. 2019Class-Balanced Loss Based on Effective Number of Samplescited by 1,239
  8. 2016End to End Learning for Self-Driving Carscited by 1,000
  9. 2019Learning Imbalanced Datasets with Label-Distribution-Aware Margin Losscited by 987
  10. 2019Decoupling Representation and Classifier for Long-Tailed Recognitioncited by 806
  11. 2018Recognition in Terra Incognitacited by 655
  12. 2017End-to-end Driving via Conditional Imitation Learningcited by 479
12 more
  1. 2018An Algorithmic Perspective on Imitation Learningcited by 455
  2. 2019BBN: Bilateral-Branch Network with Cumulative Learning for Long-Tailed Visual Recognitioncited by 403
  3. 2018ChauffeurNet: Learning to Drive by Imitating the Best and Synthesizing the Worstcited by 391
  4. 2019Learning by Cheatingcited by 361
  5. 2019Exploring the Limitations of Behavior Cloning for Autonomous Drivingcited by 343
  6. 2019Causal Confusion in Imitation Learningcited by 286
  7. 2017DART: Noise Injection for Robust Imitation Learningcited by 187
  8. 2017Deeply AggreVaTeD: Differentiable Imitation Learning for Sequential Predictioncited by 93
  9. 2020Fighting Copycat Agents in Behavioral Cloning from Observation Historiescited by 70
  10. 2017Class Rectification Hard Mining for Imbalanced Deep Learningcited by 67
  11. 2020An Evaluation of Change Point Detection Algorithmscited by 62
  12. 2019Monocular Plan View Networks for Autonomous Drivingcited by 37

Built on by

Cited by year (2026 to 7 Oct)

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

25 papers.

Sort:

By area:

  1. 2023End-to-end Autonomous Driving: Challenges and Frontierscited by 485
  2. 2022TransFuser: Imitation with Transformer-Based Sensor Fusion for Autonomous Drivingcited by 477
  3. 2024Vision-based Manipulation from Single Human Video with Open-World Object Graphscited by 46
  4. 2023A Survey on Causal Reinforcement Learningcited by 43
  5. 2022Sequence Model Imitation Learning with Unobserved Contextscited by 34
  6. 2025ContextVLA: Vision-Language-Action Model with Amortized Multi-Frame Contextcited by 34
  7. 2025Learning Long-Context Diffusion Policies via Past-Token Predictioncited by 28
  8. 2026RLDX-1 Technical Reportcited by 25
  9. 2026BPP: Long-Context Robot Imitation Learning by Focusing on Key History Framescited by 17
  10. 2022Fighting Fire with Fire: Avoiding DNN Shortcuts through Primingcited by 15
  11. 2022Resolving Copycat Problems in Visual Imitation Learning via Residual Action Predictioncited by 14
  12. 2023Domain Adaptive Imitation Learning with Visual Observationcited by 11
13 more
  1. 2024VIEW: Visual Imitation Learning with Waypointscited by 10
  2. 2026Efficient Long-Horizon Vision-Language-Action Models via Static-Dynamic Disentanglementcited by 5
  3. 2025AutoFocus-IL: VLM-based Saliency Maps for Data-Efficient Visual Imitation Learning without Extra Human Annotationscited by 3
  4. 2023Can Active Sampling Reduce Causal Confusion in Offline Reinforcement Learning?cited by 2
  5. 2024A Comprehensive Review of 3D Object Detection in Autonomous Driving: Technological Advances and Future Directionscited by 2
  6. 2025AEGIS: Human Attention-based Explainable Guidance for Intelligent Vehicle Systemscited by 2
  7. 2025A Review of Learning-Based Motion Planning: Toward a Data-Driven Optimal Control Approachcited by 1
  8. 2024Visual Imitation Learning with Calibrated Contrastive Representationcited by 0
  9. 2025End-to-end Autonomous Vehicle Following System using Monocular Fisheye Cameracited by 0
  10. 2026SkiP: When to Skip and When to Refine for Efficient Robot Manipulationcited by 0
  11. 2026SynthICL: Scalable In-context Imitation Learning with Synthetic Datacited by 0
  12. 2026Critical Interval MSE: Toward Reliable Offline Validation for Robot Manipulation Policiescited by 0
  13. 2026InfraVLA: Extending Vision-Language-Action Navigation with Infrastructure Camerascited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2021The Surprising Effectiveness of Representation Learning for Visual Imitation
  2. 2022K-VIL: Keypoints-based Visual Imitation Learning
  3. 2023Robust Visual Imitation Learning with Inverse Dynamics Representations
  4. 2025OSVI-WM: One-Shot Visual Imitation for Unseen Tasks using World-Model-Guided Trajectory Generation
  5. 2026Learning to See While Learning to Act: Diffusion Models for Active Perception in Robot Imitation
  6. 2026On the Generalization Capabilities, Design Choices and Limitations of Keypoint Imitation Learning
  7. 2021Manipulator-Independent Representations for Visual Imitation
  8. 2020A Geometric Perspective on Visual Imitation Learning