mlsanityData to 14 Oct 2026

Streaming Flow Policy: Simplifying diffusion/flow-matching policies by treating action trajectories as flow trajectories

arXiv 2505.21851 ↗First posted 28 May 2025cs.RO cs.AI cs.LGcited by 39

On the map: Diffusion models → Diffusion and flow policies for robotics

also in 1 more
Abstract (arXiv)

Recent advances in diffusion/flow-matching policies have enabled imitation learning of complex, multi-modal action trajectories. However, they are computationally expensive because they sample a trajectory of trajectories: a diffusion/flow trajectory of action trajectories. They discard intermediate action trajectories, and must wait for the sampling process to complete before any actions can be executed on the robot. We simplify diffusion/flow policies by treating action trajectories as flow trajectories. Instead of starting from pure noise, our algorithm samples from a narrow Gaussian around the last action. Then, it incrementally integrates a velocity field learned via flow matching to produce a sequence of actions that constitute a single trajectory. This enables actions to be streamed to the robot on-the-fly during the flow sampling process, and is well-suited for receding horizon policy execution. Despite streaming, our method retains the ability to model multi-modal behavior. We train flows that stabilize around demonstration trajectories to reduce distribution shift and improve imitation learning performance. Streaming flow policy outperforms prior methods while enabling faster policy execution and tighter sensorimotor loops for learning-based robot control. Project website: https://streaming-flow-policy.github.io/

Key facts. Cited by 39 papers in the map (data through 14 Oct 2026). Placed in Diffusion and flow policies for robotics, one of 14 directions in Diffusion models. Builds on 32 of its 36 references that are in the map. First posted on arXiv 28 May 2025.

Builds on

32 of 36 references.

Closest ancestors · 7

Same direction, newest first

  1. 2024ActionFlow: Equivariant, Accurate, and Efficient Policies with Spatially Symmetric Flow Matching6 Sep
  2. 2024Affordance-based Robot Manipulation with Flow Matching2 Sep
  3. 2024Streaming Diffusion Policy: Fast Policy Synthesis with Variable Noise Diffusion Models7 Jun
  4. 2024Efficient Trajectory Forecasting and Generation with Conditional Flow Matching16 Mar
  5. 2024Riemannian Flow Matching Policy for Robot Motion Learning15 Mar
  6. 2023NoMaD: Goal Masked Diffusion Policies for Navigation and Exploration11 Oct
  7. 2023Diffusion Policy: Visuomotor Policy Learning via Action Diffusion7 Mar

Other references · 25

  1. 2020Denoising Diffusion Probabilistic Modelscited by 21,076
  2. 2020Denoising Diffusion Implicit Modelscited by 9,781
  3. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,058
  4. 2022Flow Matching for Generative Modelingcited by 5,702
  5. 2024Scaling Rectified Flow Transformers for High-Resolution Image Synthesiscited by 4,181
  6. 2018Neural Ordinary Differential Equationscited by 3,779
  7. 2024π0: A Vision-Language-Action Flow Model for General Robot Controlcited by 2,407
  8. 2023Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardwarecited by 1,568
  9. 2024Octo: An Open-Source Generalist Robot Policycited by 1,484
  10. 2022Planning with Diffusion for Flexible Behavior Synthesiscited by 1,061
  11. 2024SiT: Exploring Flow and Diffusion-based Generative Models with Scalable Interpolant Transformerscited by 737
  12. 2021What Matters in Learning from Offline Human Demonstrations for Robot Manipulationcited by 725
13 more
  1. 2024Movie Gen: A Cast of Media Foundation Modelscited by 644
  2. 2018FFJORD: Free-form Continuous Dynamics for Scalable Reversible Generative Modelscited by 594
  3. 2020Neural Controlled Differential Equations for Irregular Time Seriescited by 506
  4. 2021Implicit Behavioral Cloningcited by 452
  5. 2024Pyramidal Flow Matching for Efficient Video Generative Modelingcited by 366
  6. 2023SE(3)-Stochastic Flow Matching for Protein Backbone Generationcited by 150
  7. 2023Equivariant flow matchingcited by 135
  8. 2024AlphaFold Meets Flow Matching for Generating Protein Ensemblescited by 110
  9. 2024Rolling Diffusion Modelscited by 104
  10. 2023TEDi: Temporally-Entangled Diffusion for Long-Term Motion Synthesiscited by 49
  11. 2023Provable Guarantees for Generative Behavior Cloning: Bridging Low-Level Stability and High-Level Behaviorcited by 33
  12. 2024Learning Deep Dynamical Systems using Stable Neural ODEscited by 8
  13. 2022Multiscale Sensor Fusion and Continuous Control with Neural CDEscited by 4

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 4; 2026: 35.

39 papers.

Sort:

By area:

  1. 2025Leave No Observation Behind: Real-time Correction for VLA Action Chunkscited by 46
  2. 2025SAC Flow: Sample-Efficient Reinforcement Learning of Flow-Based Policies via Velocity-Reparameterized Sequential Modelingcited by 31
  3. 2026VLA-Corrector: Lightweight Detect-and-Correct Inference for Adaptive Action Horizoncited by 15
  4. 2025UMI-on-Air: Embodiment-Aware Guidance for Embodiment-Agnostic Visuomotor Policiescited by 11
  5. 2026StreamingVLA: Streaming Vision-Language-Action Model with Action Flow Matching and Adaptive Early Observationcited by 6
  6. 2026TIDAL: Temporally Interleaved Diffusion and Action Loop for High-Frequency VLA Controlcited by 5
  7. 2025OMP: One-step Meanflow Policy with Directional Alignmentcited by 4
  8. 2026πR2: Reactive Real-time Flow Policiescited by 4
  9. 2026When Life Gives You BC, Make Q-functions: Extracting Q-values from Behavior Cloning for On-Robot Reinforcement Learningcited by 3
  10. 2026Tube Diffusion Policy: Reactive Visual-Tactile Policy Learning for Contact-rich Manipulationcited by 2
  11. 2026Diffusion Forcing Planner: History-Annealed Planning with Time-Dependent Guidance for Autonomous Drivingcited by 2
  12. 2026Where Should Action Generation Begin? A Learnable Source Prior for Generative Robot Policiescited by 2
27 more
  1. 2026TacForcing: Streaming Action Generation with Execution-Time Tactile Feedbackcited by 2
  2. 2026FlowMPC: Improving Flow Matching policies with World Modelscited by 1
  3. 2026Frequency-Aware Flow Matching for Continuous and Consistent Robotic Action Generationcited by 1
  4. 2026Flow as Flow: Modeling Robot Velocity Fields as Probability Velocity Fields for Flow-Based Object Manipulationcited by 1
  5. 2026Behavior Uncloning: Distilling Mode Redirection into Policy Weights without Inference-Time Steeringcited by 1
  6. 2026High-Fidelity One-Step Generative Visuomotor Policy via Recursive Correction, Frequency Consistency, and Contrastive Flow Matchingcited by 1
  7. 2026BICPO-VLA: Behavior-Identified Continuation Preference Optimization for Smooth Asynchronous Vision-Language-Action Controlcited by 1
  8. 2026Flow-based Policy With Distributional Reinforcement Learning in Trajectory Optimizationcited by 0
  9. 2026Recovering Hidden Reward in Diffusion-Based Policiescited by 0
  10. 2026Guided Streaming Stochastic Interpolant Policycited by 0
  11. 2026BayesFP: Posterior Estimation for Flow-Based Policies via Feynman-Kac Samplingcited by 0
  12. 2026DynaWM: A Base-VLA-Guided World Foundation Model for Moving-Object Manipulationcited by 0
  13. 2026Spatial Attention: Adapting Execution Horizons for Diffusion Policies via Observation Sensitivitycited by 0
  14. 2026Temporal Policy: History-Initialized Action Generation for Robotic Learning from Demonstrationcited by 0
  15. 2026GeoAAC: Geometry-Based Adaptive Action Chunking from Denoising Trajectories in VLA Policiescited by 0
  16. 2026Object-Centric Conditioning for Visuomotor Flow Matchingcited by 0
  17. 2026The Gaussian Is Enough: Flow-Matching Priors Do Not Help When Fine-Tuning Large Behavior Modelscited by 0
  18. 2026RAVEL: Asynchronous Rolling Inference for Flow-Based Vision-Language-Action Modelscited by 0
  19. 2026Zero-Shot Reactive Obstacle Avoidance for Generative Robot Policiescited by 0
  20. 2026Urgent Actions Go First: Urgency-Aware Denoising for Real-Time VLA Controlcited by 0
  21. 2026Variational Streaming Flow: Probabilistic Forecasting in Physical Timecited by 0
  22. 2026World-Calibrated Proposal-to-Action Flow for Vision-Language-Action Modelscited by 0
  23. 2026Safe Streaming Flow Planning by Aligning Sampling Dynamics with Execution Dynamicscited by 0
  24. 2026Adaptive Mean Flow for Responsive Closed-Loop Robot Controlcited by 0
  25. 2026Seq-Flow: Efficient Probabilistic Forecasting with Self-Rollout Error Controlcited by 0
  26. 2026Reliability-Aware Future Conditioning for Temporally Robust Robot Manipulationcited by 0
  27. 2026VITAL: From Visual Motion to Precise Actioncited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2024FlowPolicy: Enabling Fast and Robust 3D Flow-based Policy via Consistency Flow Matching for Robot Manipulation
  2. 2026Action-to-Action Flow Matching
  3. 2026PrT-Flow: Probability-Tractable Flow Matching for Reinforcement Learning
  4. 2026CoLA-Flow Policy: Temporally Coherent Imitation Learning via Continuous Latent Action Flow Matching for Robotic Manipulation
  5. 2025FlowMP: Learning Motion Fields for Robot Planning with Conditional Flow Matching
  6. 2024Learning Robotic Manipulation Policies from Point Clouds with Conditional Flow Matching
  7. 2025Flow Matching Policy Gradients
  8. 2025Fast Flow-based Visuomotor Policies via Conditional Optimal Transport Couplings