mlsanityData to 14 Oct 2026

AdaFlow: Imitation Learning with Variance-Adaptive Flow-Based Policies

arXiv 2402.04292 ↗First posted 6 Feb 2024cs.LG cs.AIcited by 50

On the map: Offline RL and imitation learning from data → Diffusion and flow policies for offline RL

Abstract (arXiv)

Diffusion-based imitation learning improves Behavioral Cloning (BC) on multi-modal decision-making, but comes at the cost of significantly slower inference due to the recursion in the diffusion process. It urges us to design efficient policy generators while keeping the ability to generate diverse actions. To address this challenge, we propose AdaFlow, an imitation learning framework based on flow-based generative modeling. AdaFlow represents the policy with state-conditioned ordinary differential equations (ODEs), which are known as probability flows. We reveal an intriguing connection between the conditional variance of their training loss and the discretization error of the ODEs. With this insight, we propose a variance-adaptive ODE solver that can adjust its step size in the inference stage, making AdaFlow an adaptive decision-maker, offering rapid inference without sacrificing diversity. Interestingly, it automatically reduces to a one-step generator when the action distribution is uni-modal. Our comprehensive empirical evaluation shows that AdaFlow achieves high performance with fast inference speed.

Key facts. Cited by 50 papers in the map (data through 14 Oct 2026). Placed in Diffusion and flow policies for offline RL, one of 13 directions in Offline RL and imitation learning from data. Builds on 50 of its 60 references that are in the map. First posted on arXiv 6 Feb 2024.

Builds on

50 of 60 references.

Closest ancestors · 3

Same direction, newest first

  1. 2023A Coupled Flow Approach to Imitation Learning29 Apr
  2. 2023IDQL: Implicit Q-Learning as an Actor-Critic Method with Diffusion Policies20 Apr
  3. 2022Is Conditional Generative Modeling all you need for Decision-Making?28 Nov

Other references · 47

  1. 2020Denoising Diffusion Probabilistic Modelscited by 21,076
  2. 2020Denoising Diffusion Implicit Modelscited by 9,781
  3. 2020Score-Based Generative Modeling through Stochastic Differential Equationscited by 9,122
  4. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,058
  5. 2022Photorealistic Text-to-Image Diffusion Models with Deep Language Understandingcited by 6,355
  6. 2022Flow Matching for Generative Modelingcited by 5,702
  7. 2019Generative Modeling by Estimating Gradients of the Data Distributioncited by 4,101
  8. 2022Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flowcited by 3,700
  9. 2022Elucidating the Design Space of Diffusion-Based Generative Modelscited by 3,045
  10. 2023Diffusion Policy: Visuomotor Policy Learning via Action Diffusioncited by 2,838
  11. 2022Video Diffusion Modelscited by 2,305
  12. 2022RT-1: Robotics Transformer for Real-World Control at Scalecited by 2,061
35 more
  1. 2022DPM-Solver: A Fast ODE Solver for Diffusion Probabilistic Model Sampling in Around 10 Stepscited by 1,979
  2. 2023Consistency Modelscited by 1,822
  3. 2016Generative Adversarial Imitation Learningcited by 1,597
  4. 2020D4RL: Datasets for Deep Data-Driven Reinforcement Learningcited by 1,427
  5. 2023LIBERO: Benchmarking Knowledge Transfer for Lifelong Robot Learningcited by 1,410
  6. 2020DiffWave: A Versatile Diffusion Model for Audio Synthesiscited by 1,354
  7. 2022Planning with Diffusion for Flexible Behavior Synthesiscited by 1,061
  8. 2022Building Normalizing Flows with Stochastic Interpolantscited by 995
  9. 2023Stochastic Interpolants: A Unifying Framework for Flows and Diffusionscited by 939
  10. 2022DPM-Solver++: Fast Solver for Guided Sampling of Diffusion Probabilistic Modelscited by 938
  11. 2021Diffusion Probabilistic Models for 3D Point Cloud Generationcited by 735
  12. 2021What Matters in Learning from Offline Human Demonstrations for Robot Manipulationcited by 725
  13. 2022Pseudo Numerical Methods for Diffusion Models on Manifoldscited by 700
  14. 2018Behavioral Cloning from Observationcited by 492
  15. 2018An Algorithmic Perspective on Imitation Learningcited by 459
  16. 2021Implicit Behavioral Cloningcited by 452
  17. 2022Analytic-DPM: an Analytic Estimate of the Optimal Reverse Variance in Diffusion Probabilistic Modelscited by 334
  18. 2022Behavior Transformers: Cloning k modes with one stonecited by 304
  19. 2022Rectified Flow: A Marginal Preserving Approach to Optimal Transportcited by 283
  20. 2023Imitating Human Behaviour with Diffusion Modelscited by 263
  21. 2023NoMaD: Goal Masked Diffusion Policies for Navigation and Explorationcited by 248
  22. 2023ViNT: A Foundation Model for Visual Navigationcited by 222
  23. 2022Diffusion-GAN: Training GANs with Diffusioncited by 205
  24. 2019Imitation Learning via Off-Policy Distribution Matchingcited by 183
  25. 2023HIVE: Harnessing Human Feedback for Instructional Visual Editingcited by 173
  26. 2023Foundation Models for Decision Making: Problems, Methods, and Opportunitiescited by 164
  27. 2022VIOLA: Imitation Learning for Vision-Based Manipulation with Object Proposal Priorscited by 136
  28. 2022Diffusion-based Molecule Generation with Informative Prior Bridgescited by 106
  29. 2022DALL-E-Bot: Introducing Web-Scale Diffusion Models to Roboticscited by 106
  30. 2021Score-Based Point Cloud Denoisingcited by 83
  31. 2022Fast Point Cloud Generation with Straight Flowscited by 56
  32. 2023VoiceFlow: Efficient Text-to-Speech with Rectified Flow Matchingcited by 49
  33. 2023Iterative α-(de)Blending: a Minimalist Deterministic Diffusion Modelcited by 46
  34. 2023Motion Flow Matching for Human Motion Synthesis and Editingcited by 38
  35. 2022IL-flOw: Imitation Learning from Observation using Normalizing Flowscited by 12

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2024: 6; 2025: 20; 2026: 24.

50 papers.

Sort:

By area:

  1. 2025GR00T N1: An Open Foundation Model for Generalist Humanoid Robotscited by 1,225
  2. 2024FlowPolicy: Enabling Fast and Robust 3D Flow-based Policy via Consistency Flow Matching for Robot Manipulationcited by 104
  3. 2024Learning Robotic Manipulation Policies from Point Clouds with Conditional Flow Matchingcited by 65
  4. 2024Affordance-based Robot Manipulation with Flow Matchingcited by 54
  5. 2025Can We Detect Failures Without Failure Data? Uncertainty-Aware Runtime Failure Detection for Imitation Learning Policiescited by 52
  6. 2025Failure Prediction at Runtime for Generative Robot Policiescited by 42
  7. 2025Much Ado About Noising: Dispelling the Myths of Generative Robotic Controlcited by 36
  8. 2024Flow Matching Imitation Learning for Multi-Support Manipulationcited by 29
  9. 2025One-Step Flow Policy Mirror Descentcited by 21
  10. 2025IMLE Policy: Fast and Sample Efficient Visuomotor Policy Learning via Implicit Maximum Likelihood Estimationcited by 13
  11. 2025FreqPolicy: Efficient Flow-based Visuomotor Policy via Frequency Consistencycited by 13
  12. 2025Fast Flow-based Visuomotor Policies via Conditional Optimal Transport Couplingscited by 9
38 more
  1. 2026One-Step Flow Policy: Self-Distillation for Fast Visuomotor Policiescited by 7
  2. 2024Local Flow Matching Generative Modelscited by 5
  3. 2025UniConFlow: A Unified Constrained Flow-Matching Framework for Certified Motion Planningcited by 5
  4. 2026Drift-Based Policy Optimization: Native One-Step Policy Learning for Online Robot Controlcited by 5
  5. 2025OMP: One-step Meanflow Policy with Directional Alignmentcited by 4
  6. 2026WarmPrior: Straightening Flow-Matching Policies with Temporal Priorscited by 4
  7. 2026HumanoidMimicGen: Data Generation for Loco-Manipulation via Whole-Body Planningcited by 4
  8. 2025Extremum Flow Matching for Offline Goal Conditioned Reinforcement Learningcited by 3
  9. 2025Train-Once Plan-Anywhere Kinodynamic Motion Planning via Diffusion Treescited by 3
  10. 2025Diffusion Bridge or Flow Matching? A Unifying Framework and Comparative Analysiscited by 3
  11. 2025Sample from What You See: Visuomotor Policy Learning via Diffusion Bridge with Observation-Embedded Stochastic Differential Equationcited by 3
  12. 2026Formalizing the Sampling Design Space of Diffusion-Based Generative Models via Adaptive Solvers and Wasserstein-Bounded Timestepscited by 3
  13. 2026ProbeFlow: Training-Free Adaptive Flow Matching for Vision-Language-Action Modelscited by 3
  14. 2025Dense-Jump Flow Matching with Non-Uniform Time Scheduling for Robotic Policies: Mitigating Multi-Step Inference Degradationcited by 2
  15. 2026Boosting Maximum Entropy Reinforcement Learning via One-Step Flow Matchingcited by 2
  16. 2026Ada3Drift: Adaptive Training-Time Drifting for One-Step 3D Visuomotor Robotic Manipulationcited by 2
  17. 2026Where Should Action Generation Begin? A Learnable Source Prior for Generative Robot Policiescited by 2
  18. 2024ReMoBot: Retrieval-Based Few-Shot Imitation Learning for Mobile Manipulation with Vision Foundation Modelscited by 1
  19. 2025GIFT: Geometry-Induced Functional Transfer for Category-level Object Manipulationcited by 1
  20. 2025DASIP: Dynamic Test-Time Compute Scaling for Robot Control with Stochastic Interpolant Policiescited by 1
  21. 2026PRISM: Performer RS-IMLE for Single-pass Multisensory Imitation Learningcited by 1
  22. 2026Accelerated Sequential Flow Matching: A Bayesian Filtering Perspectivecited by 1
  23. 2026Adaptive Time Step Flow Matching for Autonomous Driving Motion Planningcited by 1
  24. 2026Efficient Hybrid SE(3)-Equivariant Visuomotor Flow Policy via Spherical Harmonics for Robot Manipulationcited by 1
  25. 2026Agile-WAM: An Agile Tactile World Action Model for Contact-Rich Robot Controlcited by 1
  26. 2025Decision Flow Policy Optimizationcited by 0
  27. 2025Act to See, See to Act: Diffusion-Driven Perception-Action Interplay for Adaptive Policiescited by 0
  28. 2025SeFA-Policy: Fast and Accurate Visuomotor Policy Learning with Selective Flow Alignmentcited by 0
  29. 2026KAN We Flow? Advancing Robotic Manipulation with 3D Flow Matching via KAN & RWKVcited by 0
  30. 2026Primary-Fine Decoupling for Action Generation in Robotic Imitationcited by 0
  31. 2026Trajectory-Consistent Flow Matching for Robust Visuomotor Policy Learningcited by 0
  32. 2026SABER: A Scalable Action-Based Embodied Dataset for Real-World VLA Adaptationcited by 0
  33. 2026FocalPolicy: Frequency-Optimized Chunking and Locally Anchored Flow Matching for Coherent Visuomotor Policycited by 0
  34. 2026Optimal Transport Q-Learning for Flow Policy Steering and Accelerationcited by 0
  35. 2026Action Map Policy: Learning 3D Closed-loop Manipulation via Pixel Classificationcited by 0
  36. 2026CAT-Flow: Curvature-Adaptive sTeps for Flow Matchingcited by 0
  37. 2026Attention-DP3: Spatially Object-aware 3D Diffusion Policy via Geometry-aligned Attentional Conditioningcited by 0
  38. 2026Faster Visuomotor Policy Learning on Action Manifolds via Riemannian MeanFlowcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026A Survey of Flow Matching in Reinforcement Learning
  2. 2025pi-Flow: Policy-Based Few-Step Generation via Imitation Distillation
  3. 2019A Divergence Minimization Perspective on Imitation Learning Methods
  4. 2025Streaming Flow Policy: Simplifying diffusion/flow-matching policies by treating action trajectories as flow trajectories
  5. 2025Iterative Refinement of Flow Policies in Probability Space for Online Reinforcement Learning
  6. 2025Ergodic Generative Flows
  7. 2024Don't Start from Scratch: Behavioral Refinement via Interpolant-based Policy Diffusion
  8. 2026CoLA-Flow Policy: Temporally Coherent Imitation Learning via Continuous Latent Action Flow Matching for Robotic Manipulation