mlsanityData to 7 Oct 2026

A Motion-Aware Vector Quantization Framework with Centroid Reuse for Efficient VLA Inference

arXiv 2607.24148 ↗First posted 27 Jul 2026cs.AIcited by 0

On the map: Vision-language-action models → Efficient VLA inference and adaptation

Abstract (arXiv)

Vision-Language-Action (VLA) models have demonstrated strong potential for embodied AI, yet their high inference latency on GPUs limits real-time deployment. Existing accelerators, such as Dadu-Corki, improve efficiency but treat VLA models as full-precision workloads, leaving substantial redundancy in both memory and computation underexploited. In this paper, we propose VQVLA, an algorithm-hardware co-design framework that accelerates VLA inference by exploiting weight similarity and execution dynamics. We first introduce MotionVQ, a motion-aware vector quantization scheme that dynamically adjusts quantization precision based on the robot's execution state, reducing memory access while preserving task success rate. We then propose a merged-centroid vectorized GEMM paradigm that operates on the codebook-index representation, eliminating redundant multiplications through spatial aggregation and temporal reuse of centroids. To realize these optimizations, we design an accelerator that efficiently supports dynamic precision selection and centroid-reuse computation. Experimental results show that VQVLA achieves 6.5x, 2.8x, 1.9x, 3.3x, and 4.3x speedup over the A100 GPU, Dadu-Corki, LUT-DLA, CodeGEMM, and ShiftAddLLM, respectively, with negligible accuracy degradation.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Efficient VLA inference and adaptation, one of 13 directions in Vision-language-action models. Builds on 37 of its 70 references that are in the map. First posted on arXiv 27 Jul 2026.

Builds on

37 of 70 references.

References · 37

  1. 2023RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Controlcited by 2,993
  2. 2024OpenVLA: An Open-Source Vision-Language-Action Modelcited by 2,777
  3. 2024π0: A Vision-Language-Action Flow Model for General Robot Controlcited by 2,297
  4. 2023LIBERO: Benchmarking Knowledge Transfer for Lifelong Robot Learningcited by 1,338
  5. 2025GR00T N1: An Open Foundation Model for Generalist Humanoid Robotscited by 1,167
  6. 2025Fine-Tuning Vision-Language-Action Models: Optimizing Speed and Successcited by 844
  7. 2024RDT-1B: a Diffusion Foundation Model for Bimanual Manipulationcited by 670
  8. 2024QuIP#: Even Better LLM Quantization with Hadamard Incoherence and Lattice Codebookscited by 354
  9. 2023Vision-Language Foundation Models as Effective Robot Imitatorscited by 344
  10. 2024Extreme Compression of Large Language Models via Additive Quantizationcited by 223
  11. 2024ManiSkill3: GPU Parallelized Robotics Simulation and Rendering for Generalizable Embodied AIcited by 216
  12. 2023Foundation Models in Robotics: Applications, Challenges, and the Futurecited by 212
25 more
  1. 2023OWQ: Outlier-Aware Weight Quantization for Efficient Fine-Tuning and Inference of Large Language Modelscited by 113
  2. 2024PQCache: Product Quantization-based KVCache for Long Context LLM Inferencecited by 93
  3. 2023OliVe: Accelerating Large Language Models via Hardware-friendly Outlier-Victim Pair Quantizationcited by 83
  4. 2023GPT-4V(ision) for Robotics: Multimodal Task Planning from Human Demonstrationcited by 76
  5. 2024VLATest: Testing and Evaluating Vision-Language-Action Models for Robotic Manipulationcited by 43
  6. 2018XNOR Neural Engine: a Hardware Accelerator IP for 21.6 fJ/op Binary Neural Network Inferencecited by 28
  7. 2024ShiftAddLLM: Accelerating Pretrained LLMs via Post-Training Multiplication-Less Reparameterizationcited by 27
  8. 2024Fast Matrix Multiplications for Lookup Table-Quantized LLMscited by 27
  9. 2024MicroScopiQ: Accelerating Foundational Models through Outlier-Aware Microscaling Quantizationcited by 22
  10. 2020BiQGEMM: Matrix Multiplication with Lookup Table For Binary-Coding-based Quantized DNNscited by 17
  11. 2018Navion: A 2mW Fully Integrated Real-Time Visual-Inertial Odometry Accelerator for Autonomous Navigation of Nano Dronescited by 14
  12. 2024DaDu-Corki: Algorithm-Architecture Co-Design for Embodied AI-powered Robotic Manipulationcited by 12
  13. 2025VQ-LLM: High-performance Code Generation for Vector Quantization Augmented LLM Inferencecited by 12
  14. 2025LUT-DLA: Lookup Table as Efficient Extreme Low-Bit Deep Learning Acceleratorcited by 8
  15. 2025MX+: Pushing the Limits of Microscaling Formats for Efficient Large Language Model Servingcited by 8
  16. 2019eSLAM: An Energy-Efficient Accelerator for Real-Time ORB-SLAM on FPGA Platformcited by 7
  17. 2026M2XFP: A Metadata-Augmented Microscaling Data Format for Efficient Low-bit Quantizationcited by 7
  18. 2024BETA: Binarized Energy-Efficient Transformer Accelerator at the Edgecited by 6
  19. 2020Eudoxus: Characterizing and Accelerating Localization in Autonomous Machinescited by 5
  20. 2023SPADE: Sparse Pillar-based 3D Object Detection Accelerator for Autonomous Drivingcited by 4
  21. 2025Binary Weight Multi-Bit Activation Quantization for Compute-in-Memory CNN Acceleratorscited by 3
  22. 2025CodeGEMM: A Codebook-Centric Approach to Efficient GEMM in Quantized LLMscited by 3
  23. 2023Dadu-RBD: Robot Rigid Body Dynamics Accelerator with Multifunctional Pipelinescited by 2
  24. 2023VaPr: Variable-Precision Tensors to Accelerate Robot Motion Planningcited by 2
  25. 2021A Universal LiDAR SLAM Accelerator System on Low-cost FPGAcited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026DyQ-VLA: Temporal-Dynamic-Aware Quantization for Embodied Vision-Language-Action Models
  2. 2026Algorithm-Architecture Co-Design for Efficient VLA Inference via Speculative Inference and Verification
  3. 2026QVLA: Not All Channels Are Equal in Vision-Language-Action Model's Quantization
  4. 2025SQAP-VLA: A Synergistic Quantization-Aware Pruning Framework for High-Performance Vision-Language-Action Models
  5. 2026HBVLA: Pushing 1-Bit Post-Training Quantization for Vision-Language-Action Models
  6. 2025EaqVLA: Encoding-aligned Quantization for Vision-Language-Action Models
  7. 2026Deltoris: Enabling Real-time VLA Inference in Embodied AI via Bit-level Sparsity and Speculative Inference
  8. 2025PD-VLA: Accelerating Vision-Language-Action Model Integrated with Action Chunking via Parallel Decoding