mlsanityData to 14 Oct 2026

Boosting Embodied AI Agents through Perception-Generation Disaggregation and Asynchronous Pipeline Execution

arXiv 2509.09560 ↗First posted 11 Sep 2025cs.AI cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

Embodied AI systems operate in dynamic environments, requiring seamless integration of perception and generation modules to process high-frequency input and output demands. Traditional sequential computation patterns, while effective in ensuring accuracy, face significant limitations in achieving the necessary "thinking" frequency for real-world applications. In this work, we present Auras, an algorithm-system co-designed inference framework to optimize the inference frequency of embodied AI agents. Auras disaggregates the perception and generation and provides controlled pipeline parallelism for them to achieve high and stable throughput. Faced with the data staleness problem that appears when the parallelism is increased, Auras establishes a public context for perception and generation to share, thereby promising the accuracy of embodied agents. Experimental results show that Auras improves throughput by 2.54x on average while achieving 102.7% of the original accuracy, demonstrating its efficacy in overcoming the constraints of sequential computation and providing high throughput.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 51 of its 74 references that are in the map. First posted on arXiv 11 Sep 2025.

Builds on

51 of 74 references.

References · 51

  1. 2017Attention Is All You Needcited by 60,361
  2. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  3. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,399
  4. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 30,186
  5. 2020Denoising Diffusion Probabilistic Modelscited by 21,076
  6. 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,713
  7. 2023GPT-4 Technical Reportcited by 18,396
  8. 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 13,028
  9. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,058
  10. 2021Improved Denoising Diffusion Probabilistic Modelscited by 3,761
  11. 2023Sigmoid Loss for Language Image Pre-Trainingcited by 3,109
  12. 2023RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Controlcited by 3,080
38 more
  1. 2015Neural Machine Translation of Rare Words with Subword Unitscited by 2,918
  2. 2024OpenVLA: An Open-Source Vision-Language-Action Modelcited by 2,887
  3. 2023Diffusion Policy: Visuomotor Policy Learning via Action Diffusioncited by 2,838
  4. 2024π0: A Vision-Language-Action Flow Model for General Robot Controlcited by 2,407
  5. 2022RT-1: Robotics Transformer for Real-World Control at Scalecited by 2,061
  6. 2023Pythia: A Suite for Analyzing Large Language Models Across Training and Scalingcited by 1,786
  7. 2023Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardwarecited by 1,568
  8. 2023SGLang: Efficient Execution of Structured Language Model Programscited by 1,141
  9. 2022Language Models as Zero-Shot Planners: Extracting Actionable Knowledge for Embodied Agentscited by 1,035
  10. 2021What Matters in Learning from Offline Human Demonstrations for Robot Manipulationcited by 725
  11. 2023BridgeData V2: A Dataset for Robot Learning at Scalecited by 657
  12. 2018Subword Regularization: Improving Neural Network Translation Models with Multiple Subword Candidatescited by 639
  13. 2024DistServe: Disaggregating Prefill and Decoding for Goodput-optimized Large Language Model Servingcited by 617
  14. 2023Splitwise: Efficient generative LLM inference using phase splittingcited by 558
  15. 2022Socratic Models: Composing Zero-Shot Multimodal Reasoning with Languagecited by 513
  16. 2024Evaluating Real-World Robot Manipulation Policies in Simulationcited by 495
  17. 2022VIMA: General Robot Manipulation with Multimodal Promptscited by 407
  18. 2023Describe, Explain, Plan and Select: Interactive Planning with Large Language Models Enables Open-World Multi-Task Agentscited by 398
  19. 2024TinyVLA: Towards Fast, Data-Efficient Vision-Language-Action Models for Robotic Manipulationcited by 339
  20. 2017Embodied Question Answeringcited by 336
  21. 2024ReKep: Spatio-Temporal Reasoning of Relational Keypoint Constraints for Robotic Manipulationcited by 315
  22. 2023EmbodiedGPT: Vision-Language Pre-Training via Embodied Chain of Thoughtcited by 279
  23. 2024RT-H: Action Hierarchies Using Languagecited by 210
  24. 2022Stochastic Trajectory Prediction via Motion Indeterminacy Diffusioncited by 185
  25. 2023Distilled Feature Fields Enable Few-Shot Language-Guided Manipulationcited by 130
  26. 2024LLaVA-Phi: Efficient Multi-Modal Assistant with Small Language Modelcited by 129
  27. 2025Dispider: Enabling Video LLMs with Active Real-Time Interaction via Disentangled Perception, Decision, and Reactioncited by 127
  28. 2022DALL-E-Bot: Introducing Web-Scale Diffusion Models to Roboticscited by 106
  29. 2024DistriFusion: Distributed Parallel Inference for High-Resolution Diffusion Modelscited by 98
  30. 2024Cobra: Extending Mamba to Multi-Modal Large Language Model for Efficient Inferencecited by 97
  31. 2024Hierarchical Diffusion Policy for Kinematics-Aware Multi-Task Robotic Manipulationcited by 93
  32. 2023CIRCLE: Capture In Rich Contextual Environmentscited by 79
  33. 2024Humanoid Locomotion as Next Token Predictioncited by 69
  34. 2024MemServe: Context Caching for Disaggregated LLM Serving with Elastic Memory Poolcited by 68
  35. 2024xDiT: an Inference Engine for Diffusion Transformers (DiTs) with Massive Parallelismcited by 56
  36. 2024In-Context Learning Enables Robot Action Prediction in LLMscited by 27
  37. 2024PipeInfer: Accelerating LLM Inference using Asynchronous Pipelined Speculationcited by 14
  38. 2024Multi-Bin Batching for Increasing LLM Inference Throughputcited by 9

50 most cited of 51 All 51 (JSON)

Built on by

Not cited yet.