mlsanityData to 7 Oct 2026

MotionStreamer: Streaming Motion Generation via Diffusion-based Autoregressive Model in Causal Latent Space

arXiv 2503.15451 ↗First posted 19 Mar 2025cs.CVcited by 82

Not on the map. Why?

Abstract (arXiv)

This paper addresses the challenge of text-conditioned streaming motion generation, which requires us to predict the next-step human pose based on variable-length historical motions and incoming texts. Existing methods struggle to achieve streaming motion generation, e.g., diffusion models are constrained by pre-defined motion lengths, while GPT-based methods suffer from delayed response and error accumulation problem due to discretized non-causal tokenization. To solve these problems, we propose MotionStreamer, a novel framework that incorporates a continuous causal latent space into a probabilistic autoregressive model. The continuous latents mitigate information loss caused by discretization and effectively reduce error accumulation during long-term autoregressive generation. In addition, by establishing temporal causal dependencies between current and historical motion latents, our model fully utilizes the available information to achieve accurate online motion decoding. Experiments show that our method outperforms existing approaches while offering more applications, including multi-round generation, long-term generation, and dynamic motion composition. Project Page: https://zju3dv.github.io/MotionStreamer/

Key facts. Cited by 82 papers in the map (data through 7 Oct 2026). Builds on 63 of its 83 references that are in the map. First posted on arXiv 19 Mar 2025.

Builds on

63 of 83 references.

References · 63

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,204
  3. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,456
  4. 2020Denoising Diffusion Probabilistic Modelscited by 20,851
  5. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,694
  6. 2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,491
  7. 2023LLaMA: Open and Efficient Foundation Language Modelscited by 13,958
  8. 2022Scalable Diffusion Models with Transformerscited by 5,931
  9. 2022Classifier-Free Diffusion Guidancecited by 5,717
  10. 2017Neural Discrete Representation Learningcited by 4,650
  11. 2021RoFormer: Enhanced Transformer with Rotary Position Embeddingcited by 4,581
  12. 2019AMASS: Archive of Motion Capture as Surface Shapescited by 1,208
38 more
  1. 2022MaskGIT: Masked Generative Image Transformercited by 1,082
  2. 2022Human Motion Diffusion Modelcited by 1,062
  3. 2022MotionDiffuse: Text-Driven Human Motion Generation with Diffusion Modelcited by 768
  4. 2022Autoregressive Image Generation using Residual Quantizationcited by 706
  5. 2016Keep it SMPL: Automatic Estimation of 3D Human Pose and Shape from a Single Imagecited by 692
  6. 2024Autoregressive Image Generation without Vector Quantizationcited by 644
  7. 2023Language Model Beats Diffusion -- Tokenizer is Key to Visual Generationcited by 633
  8. 2023T2M-GPT: Generating Human Motion from Textual Descriptions with Discrete Representationscited by 577
  9. 2022Executing your Commands via Motion Diffusion in Latent Spacecited by 567
  10. 2023MultiDiffusion: Fusing Diffusion Paths for Controlled Image Generationcited by 529
  11. 2023MotionGPT: Human Motion as a Foreign Languagecited by 521
  12. 2022TEMOS: Generating diverse human motions from textual descriptionscited by 486
  13. 2023MoMask: Generative Masked Modeling of 3D Human Motionscited by 388
  14. 2022TM2T: Stochastic and Tokenized Modeling for the Reciprocal Generation of 3D Human Motions and Textscited by 357
  15. 2023Human Motion Diffusion as a Generative Priorcited by 354
  16. 2020Query-Key Normalization for Transformerscited by 266
  17. 2021BABEL: Bodies, Action and Behavior with English Labelscited by 261
  18. 2023TMR: Text-to-Motion Retrieval Using Contrastive 3D Human Motion Synthesiscited by 194
  19. 2022TEACH: Temporal Action Composition for 3D Humanscited by 173
  20. 2024MotionLCM: Real-time Controllable Motion Generation via Latent Consistency Modelcited by 164
  21. 2023MotionGPT: Finetuned LLMs Are General-Purpose Motion Generatorscited by 145
  22. 2023MMM: Generative Masked Motion Modelcited by 129
  23. 2023AttT2M: Text-Driven Human Motion Generation with Multi-Perspective Attention Mechanismcited by 116
  24. 2022Why Exposure Bias Matters: An Imitation Learning Perspective of Error Accumulation in Language Generationcited by 106
  25. 2023HumanTOMATO: Text-aligned Whole-body Motion Generationcited by 103
  26. 2024CLoSD: Closing the Loop between Simulation and Diffusion for multi-task character controlcited by 103
  27. 2024MotionLLM: Understanding Human Behaviors from Human Motions and Videoscited by 89
  28. 2023HumanMAC: Masked Motion Completion for Human Motion Predictioncited by 81
  29. 2024DartControl: A Diffusion-Based Autoregressive Motion Model for Real-Time Text-Driven Motion Controlcited by 80
  30. 2023Fg-T2M: Fine-Grained Text-Driven Human Motion Generation via Diffusion Modelcited by 79
  31. 2024Rethinking Diffusion for Text-Driven Human Motion Generation: Redundant Representations, Evaluation, and Masked Autoregressioncited by 78
  32. 2024Seamless Human Motion Composition with Blended Positional Encodingscited by 77
  33. 2023Priority-Centric Human Motion Generation in Discrete Latent Spacecited by 66
  34. 2024Autoregressive Speech Synthesis without Vector Quantizationcited by 65
  35. 2023AvatarGPT: All-in-One Framework for Motion Understanding, Planning, Generation and Beyondcited by 63
  36. 2020Simple and Effective VAE Training with Calibrated Decoderscited by 57
  37. 2024Taming Diffusion Probabilistic Models for Character Controlcited by 56
  38. 2024BAMM: Bidirectional Autoregressive Motion Modelcited by 54

50 most cited of 63 All 63 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 21; 2026: 61.

82 papers.

Sort:

By area:

  1. 2025Go to Zero: Towards Zero-shot Motion Generation with Million-scale Datacited by 67
  2. 2025TokenHSI: Unified Synthesis of Physical Human-Scene Interactions through Task Tokenizationcited by 46
  3. 2025HY-Motion 1.0: Scaling Flow Matching Models for Text-To-Motion Generationcited by 32
  4. 2025Absolute Coordinates Make Motion Generation Easycited by 27
  5. 2025SkillMimic-V2: Learning Robust and Generalizable Interaction Skills from Sparse and Noisy Demonstrationscited by 23
  6. 2025From Language to Locomotion: Retargeting-free Humanoid Control via Motion Latent Guidancecited by 17
  7. 2025FloodDiffusion: Tailored Diffusion Forcing for Streaming Motion Generationcited by 14
  8. 2026TextOp: Real-time Interactive Text-Driven Humanoid Robot Motion Generation and Controlcited by 14
  9. 2025SIGMAN:Scaling 3D Human Gaussian Generation with Millions of Assetscited by 12
  10. 2025CoDA: Coordinated Diffusion Noise Optimization for Whole-Body Manipulation of Articulated Objectscited by 10
  11. 2026LLaMo: Scaling Pretrained Language Models for Unified Motion Understanding and Generation with Continuous Autoregressive Tokenscited by 10
  12. 2025Text-driven Motion Generation: Overview, Challenges and Directionscited by 8
38 more
  1. 2025Pulp Motion: Framing-aware multimodal camera and human motion generationcited by 8
  2. 2025Towards Interactive Intelligence for Digital Humanscited by 8
  3. 2026Causal Motion Diffusion Models for Autoregressive Motion Generationcited by 7
  4. 2025Interact2Ar: Full-Body Human-Human Interaction Generation via Autoregressive Diffusion Modelscited by 6
  5. 2026UMO: Unified In-Context Learning Unlocks Motion Foundation Model Priorscited by 5
  6. 2026Heracles: Bridging Precise Tracking and Generative Synthesis for General Humanoid Controlcited by 5
  7. 2026LottieGPT: Tokenizing Vector Animation for Autoregressive Generationcited by 5
  8. 2026CoMoVi: Co-Generation of 3D Human Motions and Realistic Videoscited by 4
  9. 2026BiMotion: B-spline Motion for Text-guided Dynamic 3D Character Generationcited by 4
  10. 2026ActionPlan: Future-Aware Streaming Motion Synthesis via Frame-Level Action Planningcited by 4
  11. 2025X-MoGen: Unified Motion Generation across Humans and Animalscited by 3
  12. 2025Think Before You Move: Latent Motion Reasoning for Text-to-Motion Generationcited by 3
  13. 2026Gloria: Consistent Character Video Generation via Content Anchorscited by 3
  14. 2026MotionHiFlow: Text-to-motion via hierarchical flow matchingcited by 3
  15. 2025MoLingo: Motion-Language Alignment for Text-to-Human Motion Generationcited by 2
  16. 2026ProAct: Harnessing Streaming Motion Generation and Agentic Reasoning for Real-Time Embodied Social Interactioncited by 2
  17. 2026PRISM: Streaming Human Motion Generation with Per-Joint Latent Decompositioncited by 2
  18. 2026Riemannian Motion Generation: A Unified Framework for Human Motion Representation and Generation via Riemannian Flow Matchingcited by 2
  19. 2026HandX: Scaling Bimanual Motion and Interaction Generationcited by 2
  20. 2025ReactDance: Hierarchical Representation for High-Fidelity and Coherent Long-Form Reactive Dance Generationcited by 1
  21. 2025InterMoE: Individual-Specific 3D Human Interaction Generation via Dynamic Temporal-Selective MoEcited by 1
  22. 2026PhysMoDPO: Physically-Plausible Humanoid Motion with Preference Optimizationcited by 1
  23. 2026ReactMotion: Generating Reactive Listener Motions from Speaker Utterancecited by 1
  24. 2026Coordinate-Based Dual-Constrained Autoregressive Motion Generationcited by 1
  25. 2026Seeing Without Eyes: 4D Human-Scene Understanding from Wearable IMUscited by 1
  26. 2026IAM: Identity-Aware Human Motion and Shape Joint Generationcited by 1
  27. 2026UMo: Unified Sparse Motion Modeling for Real-Time Co-Speech Avatarscited by 1
  28. 2026MIND: Multi-Scale Intent Diffusion for Text-Driven Physics-Based Humanoid Controlcited by 1
  29. 2026EchoAvatar: Real-time Generative Avatar Animation from Audio Streamscited by 1
  30. 2026AnyMo: Scaling Any-Modality Conditional Motion Generation with Masked Modelingcited by 1
  31. 2026ReactiveBFM: Reactive Closed-Loop Motion Planning Towards Universal Humanoid Whole-Body Controlcited by 1
  32. 2026GIRAF: Towards Generalizable Human Interactions with Articulated Objectscited by 1
  33. 2026Surface Keypoint Representation for Multi-Object and Articulated Human-Object Interaction Generationcited by 1
  34. 2026Puppeteer: Object-Grounded Posture-Aware Co-Speech Gesture Generationcited by 1
  35. 2025LILAC: Long-sequence Incremental Low-latency Arbitrary Motion Stylization via Streaming VAE-Diffusion with Causal Decodingcited by 0
  36. 2025Endowing GPT-4 with a Humanoid Body: Building the Bridge Between Off-the-Shelf VLMs and the Physical Worldcited by 0
  37. 2025Robust Motion Generation using Part-level Reliable Data from Videoscited by 0
  38. 2026TriC-Motion: Tri-Domain Causal Modeling Grounded Text-to-Motion Generationcited by 0

50 most cited of 82 All 82 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2023AMD: Autoregressive Motion Diffusion
  2. 2026SoulX-LiveAct: Towards Hour-Scale Real-Time Human Animation with Neighbor Forcing and ConvKV Memory
  3. 2023DiffusionPhase: Motion Diffusion in Frequency Domain
  4. 2026From Diffusion to Flow: Efficient Motion Generation in MotionGPT3
  5. 2025MotionStream: Real-Time Video Generation with Interactive Motion Controls
  6. 2024RDM: Recurrent Diffusion Model for Human Motion Generation
  7. 2024MotionMix: Weakly-Supervised Diffusion for Controllable Motion Generation
  8. 2023Controllable Motion Synthesis and Reconstruction with Autoregressive Diffusion Models