mlsanityData to 7 Oct 2026

WA-SpecDec: World-Aware Speculative Decoding for Vision-Language-Action Models

arXiv 2608.08725 ↗First posted 9 Aug 2026cs.ROcited by 0

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

Abstract (arXiv)

Vision-language-action (VLA) policies generate robot controls autoregressively, making closed-loop latency dominated by repeated target-model forward passes. Speculative decoding reduces this cost by verifying blocks of draft action tokens in parallel, and recent VLA methods further relax token-level acceptance because small differences in action-token space often map to similar continuous controls. However, this relaxation remains scene-agnostic. A fixed token-distance tolerance treats the same action-token deviation as equally safe across states, although deviations that are harmless in free space can cause collisions or grasp failures near contact. We propose WA-SpecDec, a world-aware speculative decoding framework that injects world-model-derived physical scene awareness during the VLA prefill stage, producing shared world-aware prefill states for draft proposal and target verification without changing the relaxed acceptance rule. Across three state-of-the-art relaxed acceptance schemes, WA-SpecDec preserves higher task success under looser relaxation and enables longer accepted prefixes. At comparable-success operating points, WA-SpecDec achieves a 1.5x matched-success speedup over VLA speculative decoding alone and reduces near-contact failure (NCF) by 18.6% on average relative to the corresponding speculative baselines.

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 30 of its 38 references that are in the map. First posted on arXiv 9 Aug 2026.

Builds on

30 of 38 references.

Closest ancestors · 3

Same direction, newest first

  1. 2026HeiSD: Hybrid Speculative Decoding for Embodied Vision-Language-Action Models with Kinematic Awareness18 Mar
  2. 2026KERV: Kinematic-Rectified Speculative Decoding for Embodied VLA Models2 Mar
  3. 2025Spec-VLA: Speculative Decoding for Vision-Language-Action Models with Relaxed Acceptance30 Jul

Other references · 27

  1. 2023Sigmoid Loss for Language Image Pre-Trainingcited by 3,048
  2. 2024OpenVLA: An Open-Source Vision-Language-Action Modelcited by 2,777
  3. 2025Wan: Open and Advanced Large-Scale Video Generative Modelscited by 2,536
  4. 2022RT-1: Robotics Transformer for Real-World Control at Scalecited by 2,007
  5. 2022Fast Inference from Transformers via Speculative Decodingcited by 1,615
  6. 2024Octo: An Open-Source Generalist Robot Policycited by 1,427
  7. 2023LIBERO: Benchmarking Knowledge Transfer for Lifelong Robot Learningcited by 1,338
  8. 2018Learning Latent Dynamics for Planning from Pixelscited by 1,221
  9. 2023Accelerating Large Language Model Decoding with Speculative Samplingcited by 975
  10. 2023Open X-Embodiment: Robotic Learning Datasets and RT-X Modelscited by 849
  11. 2024Medusa: Simple LLM Inference Acceleration Framework with Multiple Decoding Headscited by 821
  12. 2018Recurrent World Models Facilitate Policy Evolutioncited by 761
15 more
  1. 2022R3M: A Universal Visual Representation for Robot Manipulationcited by 678
  2. 2024EAGLE: Speculative Sampling Requires Rethinking Feature Uncertaintycited by 564
  3. 2024Revisiting Feature Prediction for Learning Visual Representations from Videocited by 490
  4. 2024Evaluating Real-World Robot Manipulation Policies in Simulationcited by 484
  5. 2023TD-MPC2: Scalable, Robust World Models for Continuous Controlcited by 482
  6. 2025UniVLA: Learning to Act Anywhere with Task-centric Latent Actionscited by 386
  7. 2023SpecInfer: Accelerating Generative Large Language Model Serving with Tree-based Speculative Inference and Verificationcited by 359
  8. 2022Real-World Robot Learning with Masked Visual Pre-trainingcited by 241
  9. 2023Where are we in the search for an Artificial Visual Cortex for Embodied Intelligence?cited by 215
  10. 2023Speculative Decoding with Big Little Decodercited by 163
  11. 2023Language-Driven Representation Learning for Roboticscited by 163
  12. 2023REST: Retrieval-Based Speculative Decodingcited by 148
  13. 2022Global Planning for Contact-Rich Manipulation via Local Smoothing of Quasi-dynamic Contact Modelscited by 46
  14. 2021Tactile-RL for Insertion: Generalization to Objects of Unknown Geometrycited by 45
  15. 2021A Survey of Robot Manipulation in Contactcited by 26

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026FlashVLA: Streaming Action Decoding for Fast and Asynchronous VLA Inference
  2. 2025Think Twice, Act Once: Token-Aware Compression and Action Reuse for Efficient Inference in Vision-Language-Action Models
  3. 2026World-Coherent Decoding: Self-Verifying Test-Time Planning for World Action Models
  4. 2025PD-VLA: Accelerating Vision-Language-Action Model Integrated with Action Chunking via Parallel Decoding
  5. 2026MMSpec: Benchmarking Speculative Decoding for Vision-Language Models
  6. 2026SA-VLA: State-aware tokenizer for improving Vision-Language-Action Models' performance
  7. 2025NanoVLA: Routing Decoupled Vision-Language Understanding for Nano-sized Generalist Robotic Policies
  8. 2026A1: A Fully Transparent Open-Source, Adaptive and Efficient Truncated Vision-Language-Action Model