mlsanityData to 7 Oct 2026

Diffusion and flow policies for robotics

1,490 papers14 this weekMedium confidence in this grouping: mostly one topic, with blurry edgesMedium confidence in this grouping: mostly one topic, with blurry edges · About

Keywords: policy, diffusion, policies, action, trajectory

Key facts. Diffusion and flow policies for robotics: 1,490 papers on arXiv since 2022 (data through 7 Oct 2026). 616 of its 1,490 mapped papers (41%) cite “Diffusion Policy: Visuomotor Policy Learning via Action Diffusion” (2023, arXiv 2303.04137). 252 of its 1,490 mapped papers (17%) cite “Training Diffusion Models with Reinforcement Learning” (2023, arXiv 2305.13301). 81 new papers 17 Sep–7 Oct.

What’s new

17 Sep – 7 Oct 2026

81 new papers, 17 Sep – 7 Oct; 14 this week.

  1. This week · 14
  2. Plan-and-Patch: Diffusion Language Models for Agentic Planning
  3. RoboPrompt: Intuitive Robot Policy Steering with Sparse Human Input
  4. RFPO: Rectified Flow Policy Optimization for Embodied Control
  5. Immiscible Diffusion Policy: Preserving Multimodal Robot Actions through Label-Free Noise Assignment
  6. Constrained Diffusion for Data-Scarce Orbital Monte Carlo in Constellation Tasking
  7. StairVLA: Stage-Aware Hierarchical Action Generation for Vision-Language-Action Models
  8. ESP: Energy-Score Policy for One-Step Multimodal Action Generation
  9. Internalizing Agent Experience into Diffusion Model Weights via On-Policy Context Distillation also fits: Image generation, text-to-image and editing
  10. P3: Persistent Particle Planning for Constrained Diffusion Control
  11. Reachability-Aware Diffusion Policy Optimization
All 81
  1. Bilinear Flow Policy: Distributional Extrapolation for Goal-Conditioned Visuomotor Imitation
  2. Demonstration-Calibrated Port-Hamiltonian Retuning for Manipulation Policies
  3. The Unexpired Plan: A Free Monitor for Accelerated Diffusion Policies
  4. Spacecraft Rendezvous Trajectory Generation with Modular Constraints via Diffusion Model Composition
  5. Earlier · 67
  6. Identifiable World Models from Pretrained Diffusion Representations
  7. Learning Conditional Source Distribution via Flow Reversal for Temporal Flow Matching
  8. Diffusion Transformers are Provably Optimal In-context Generators also fits: Diffusion foundations and sampling
  9. MGPO: Manifold-Guided Diffusion Alignment for Task-Aware Dataset Distillation
  10. Learning Task and Motion Plans from Real Demonstrations with Hybrid Flow Matching
  11. Score-Calibrated Flow for Sampling from Unnormalized Densities with Applications to Generative Online Reinforcement Learning
  12. FASTER: Fast Adjoint Stochastic Transport for Endpoint Refinement in Reward-Guided Image Editing also fits: Image generation, text-to-image and editing
  13. Native Action-Prior Learning from Videos for World Action Models
  14. Equivariant Visual-Tactile Diffusion Policy for Contact-Rich Manipulation
  15. Safe Streaming Flow Planning by Aligning Sampling Dynamics with Execution Dynamics
  16. CriticHack: Evaluating Visual Rewards Under Robot Policy Optimization
  17. iADD: Improving Alignment and Diversity in Diffusion Policy Optimization
  18. Flow Matching Reinforcement for 3D Mesh Generation via Dynamic Homing Optimization
  19. Kinematic MeanFlow: One-Step Action Generation Policy for Robotic Foundation Models also fits: Flow matching, time series and science
  20. Diffusion Policy Improvement with Proposal-Conditioned Refinement Flows
  21. Discrete Forcing: Infusing Discrete Guidance into Continuous Denoising for Few-Step Action Experts
  22. DiFF: Doppler-informed Flow Matching for Human Motion Flow also fits: 3D, point cloud and scene generation
  23. Cue the Flow: Steering Flow-Matching Policies for Open-World Delivery Manipulation
  24. Loop-Free Inverse Reinforcement Learning via Sequential Value Recovery with Q-Score Matching
  25. Online Evolution Strategy for Flow-Matching VLA Policies via Self-Supervised Trajectory Distribution Optimization
  26. Conditional Generation of Creative Chess Puzzles with Diffusion Models
  27. Multi-Agent Flow Matching with Decoupled Generative Guidance
  28. PhysWAM: Physically Consistent World Action Model for Autonomous Driving
  29. FlowMap-OPD: Rollout--Kernel Separation for On-Policy Distillation of Few-Step Flow-Map Generators
  30. Urgent Actions Go First: Urgency-Aware Denoising for Real-Time VLA Control
  31. DROM: A Language-Guided Diffusion Framework for Multi-Skill Robotic Manipulation
  32. PreferenceFlow: Test-Time Guidance of Flow-Matching Robot Policies from Human Interventions
  33. PE-OPSD: Internalizing Prompt Enhancement into Flow-matching Models via On-Policy Self-Distillation also fits: Image generation, text-to-image and editing
  34. EquivDP3: A SIM(3)-Invariant Point-Cloud Encoder for Data-Efficient Humanoid Loco-Manipulation
  35. Trajectory-Level Mode Guidance for Controllable Diffusion-Based Multi-Robot Motion Planning
  36. Online Versatile Incremental Learning: Towards Class and Domain-Agnostic Adaptation at Any Time also fits: Image generation, text-to-image and editing
  37. Zero-Shot Reactive Obstacle Avoidance for Generative Robot Policies
  38. Alignment-Guided Flow Transformer for Efficient Vision-Language-Action Policy Learning
  39. Rethinking Causal Action Tokenization with Conditional Annealing in Flow Matching
  40. Spatial Grafting: Grounding 3D Features for Flow-Matching Robot Policies
  41. QAMM: Adjoint MeanFlow Matching for Few-Step Offline Reinforcement Learning
  42. Quantile Head for Vision-Language-Action Models
  43. Principal Steering Subspaces for Online Adaptation of Frozen Generative Robot Policies
  44. Steer2Grasp: Inference-Time Embodiment-Aware Steering for Diverse Physically Feasible Grasp Diffusion
  45. Safe Score Matching: Diffusion Policies with Hamilton-Jacobi Reachability for Online Safe Reinforcement Learning
  46. Schrödinger--Föllmer Actor--Critic: Diffusion Policy Improvement with Finite-Sample Analysis
  47. Notes on Generative Modeling for Feedback Control and Planning
  48. Optimizing H-Graph Hybridization for Diffusion-Guided RRT
  49. Plan-to-Synthesis: Cross-City Human Mobility Generation via Semantic Latent Flow Matching
  50. Timestep Weighting: A Hidden Key to Effective ELBO-Based Flow-Matching RL
  51. RefAdapt-DiT: Adaptive Joint Attention for Reference-Conditioned Diffusion Transformers also fits: Diffusion foundations and sampling
  52. Skeletons in Flow: Graph Structured Flow Matching for Human Motion Prediction also fits: Video generation
  53. SoGuDiff: Socially Guided Diffusion for Steerable, Norm-Grounded Robot Navigation
  54. Policy-Calibrated DAgger: Offline Calibrated Noise Injection for Imitation Learning
  55. Faster Visuomotor Policy Learning on Action Manifolds via Riemannian MeanFlow
  56. MorphIK: Morphology-Conditioned Neural Inverse Kinematics for Unknown Robots
  57. Decoupled Early Exits for Task-Dependent Compute Allocation in Flow-Matching VLAs
  58. Uncertainty-Gated Exploration Noise Suppresses Task Collapse in Online RL Fine-Tuning of a Flow-Matching Vision-Language-Action Policy
  59. Beyond Future Prediction: Denoising as Generative Adaptation for Robot Control
  60. Distillation for Efficient Multitask Manipulation Policies via Conditional Flow Matching
  61. The Gaussian Is Enough: Flow-Matching Priors Do Not Help When Fine-Tuning Large Behavior Models
  62. History-Conditioned Flow Matching for Probabilistic Dynamics of Tendon-Driven Continuum Robots
  63. JAMB: Joint Action-Motion Diffusion for Bimanual Manipulation
  64. PredActor: Predictive Action Diffusion for Steerable Onboard Humanoid Control
  65. Object-Centric Conditioning for Visuomotor Flow Matching
  66. CARF: Contrastive Attraction-Repulsion of Failure-Guided Flow Matching
  67. AcousticDiffusion: Semantically Conditioned Audio-Guided Diffusion Policy for Search-and-Rescue Assistance also fits: Audio, speech and music
  68. NaViRrator: Robot Navigation from Human-Readable Maps through a Learned Visual Route
  69. Fewer Steps, Better Actions: Rethinking Flow-Matching Inference for VLA Policies
  70. Learning Foresight without Explicit Trajectories for 3D Diffusion Policies
  71. TraceFlow: Guiding Frozen Flow-Matching Robot Policies with Success and Failure Traces
  72. LYRIC: Language-Driven Physics-Based Character Control for Contact-Rich Whole-Body Object Interaction

Foundations

Most cited by papers here

  1. Diffusion Policy: Visuomotor Policy Learning via Action Diffusion2023cited by 616 here
  2. Training Diffusion Models with Reinforcement Learning2023cited by 252 here
  3. Diffusion Policies as an Expressive Policy Class for Offline Reinforcement Learning2022cited by 237 here
  4. Is Conditional Generative Modeling all you need for Decision-Making?2022cited by 228 here
  5. Diffusion Model Alignment Using Direct Preference Optimization2023cited by 182 here

Lineage

The papers this direction builds on, oldest first.

Each paper on the line cites the one before.

  1. 2020Denoising Diffusion Probabilistic Modelscited by 999 here
  2. 2020Denoising Diffusion Implicit Modelscited by 519 here
  3. 2021Improved Denoising Diffusion Probabilistic Modelscited by 152 here
  4. 2021Diffusion Models Beat GANs on Image Synthesiscited by 316 here
  5. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 458 here
  6. 2022Hierarchical Text-Conditional Image Generation with CLIP Latentscited by 142 here
  7. 2022Photorealistic Text-to-Image Diffusion Models with Deep Language Understandingcited by 177 here
  8. 2022Is Conditional Generative Modeling all you need for Decision-Making?cited by 228 here
  9. 2023Imitating Human Behaviour with Diffusion Modelscited by 125 here
  10. 2023IDQL: Implicit Q-Learning as an Actor-Critic Method with Diffusion Policiescited by 137 here
  11. 2023Training Diffusion Models with Reinforcement Learningcited by 252 here
  12. 2023Directly Fine-Tuning Diffusion Models on Differentiable Rewardscited by 134 here
12 more it builds on, not in this chain
  1. 2020Score-Based Generative Modeling through Stochastic Differential Equationscited by 462 here
  2. 2022Elucidating the Design Space of Diffusion-Based Generative Modelscited by 139 here
  3. 2022Classifier-Free Diffusion Guidancecited by 276 here
  4. 2022Diffusion Policies as an Expressive Policy Class for Offline Reinforcement Learningcited by 237 here
  5. 2022Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flowcited by 218 here
  6. 2022Flow Matching for Generative Modelingcited by 359 here
  7. 2022Scalable Diffusion Models with Transformerscited by 195 here
  8. 2023Consistency Modelscited by 130 here
  9. 2023Diffusion Policy: Visuomotor Policy Learning via Action Diffusioncited by 616 here
  10. 2023SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesiscited by 127 here
  11. 2023Diffusion Model Alignment Using Direct Preference Optimizationcited by 182 here
  12. 2024Scaling Rectified Flow Transformers for High-Resolution Image Synthesiscited by 174 here

The 24 most cited of 2,735.

Browse

30 of 1,490, ranked 6–35

  1. 6IDQL: Implicit Q-Learning as an Actor-Critic Method with Diffusion Policies2023137
  2. 7Directly Fine-Tuning Diffusion Models on Differentiable Rewards2023134
  3. 8Imitating Human Behaviour with Diffusion Models2023125
  4. 9Goal-Conditioned Imitation Learning using Score-based Diffusion Policies2023125
  5. 10DPOK: Reinforcement Learning for Fine-tuning Text-to-Image Diffusion Models2023125
  6. 11Flow-GRPO: Training Flow Matching Models via Online RL2025121
  7. 12Diffusion Policy Policy Optimization2024117
  8. 13Using Human Feedback to Fine-tune Diffusion Models without Any Reward Model2023101
  9. 14Aligning Text-to-Image Diffusion Models with Reward Backpropagation202398
  10. 15Motion Planning Diffusion: Learning and Planning of Robot Motions with Diffusion Models202395
All 30
  1. 16RDT-1B: a Diffusion Foundation Model for Bimanual Manipulation202489
  2. 17Consistency Policy: Accelerated Visuomotor Policies via Consistency Distillation202485
  3. 18DanceGRPO: Unleashing GRPO on Visual Generation202584
  4. 193D Diffuser Actor: Policy Diffusion with 3D Scene Representations202479
  5. 20AdaptDiffuser: Diffusion Models as Adaptive Self-evolving Planners202378
  6. 21Efficient Diffusion Policies for Offline Reinforcement Learning202378
  7. 22NoMaD: Goal Masked Diffusion Policies for Navigation and Exploration202370
  8. 23Contrastive Energy Prediction for Exact Energy-Guided Diffusion Sampling in Offline Reinforcement Learning202369
  9. 24Learning a Diffusion Model Policy from Rewards via Q-Score Matching202369
  10. 25Policy Representation via Diffusion Probability Model for Reinforcement Learning202361
  11. 26SE(3)-DiffusionFields: Learning smooth cost functions for joint grasp and motion optimization through diffusion202261
  12. 27MixGRPO: Unlocking Flow-based GRPO Efficiency with Mixed ODE-SDE202558
  13. 28Guided Conditional Diffusion for Controllable Traffic Simulation202258
  14. 29Flow Q-Learning202552
  15. 30FlowPolicy: Enabling Fast and Robust 3D Flow-based Policy via Consistency Flow Matching for Robot Manipulation202452
  16. 31Hierarchical Diffusion Policy for Kinematics-Aware Multi-Task Robotic Manipulation202451
  17. 32Diffusion-based Reinforcement Learning via Q-weighted Variational Policy Optimization202450
  18. 33Adjoint Matching: Fine-tuning Flow and Diffusion Generative Models with Memoryless Stochastic Optimal Control202448
  19. 34Human Preference Score: Better Aligning Text-to-Image Models with Human Preference202347
  20. 35DiffusionNFT: Online Diffusion Reinforcement with Forward Process202546