mlsanityData to 7 Oct 2026

VLA-ACL: Action-Consistent Visual Token Pruning for Efficient Vision-Language-Action Models

arXiv 2610.08133 ↗First posted 6 Oct 2026cs.RO cs.CVcited by 0

On the map: Vision-language-action models → Efficient VLA inference and adaptation also fits: Generalist VLA models and benchmarks

also in 1 more
Abstract (arXiv)

Vision-Language-Action (VLA) models achieve strong robotic manipulation performance but incur high computational costs from processing long token sequences at every control step, limiting real-time deployment. Visual token pruning offers a direct solution, as visual patches dominate the input sequence and contain considerable redundancy. Existing approaches, however, either rely on indirect training-free heuristics, such as attention scores and motion thresholds, or require costly fine-tuning of the base VLA model. We introduce VLA-ACL (Action Consistency Learning), which learns a lightweight visual token pruning policy through action-level supervision while keeping the base VLA model entirely frozen. The training objective encourages actions produced from pruned visual contexts to remain consistent with the full-context teacher, with ground-truth actions as auxiliary supervision. This directly ties token selection to its effect on the downstream control output. Experiments on LIBERO and real-world manipulation tasks show that VLA-ACL prunes up to 87.5% of visual tokens while retaining competitive performance, reduces computation by up to 75%, and achieves a 1.5x inference speedup. These results establish a stronger performance-efficiency trade-off than existing frozen-VLA pruning methods and demonstrate the value of action-level supervision for visual token selection. Code is available at https://github.com/du-owen/VLA-ACL.

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 34 of its 38 references that are in the map. First posted on arXiv 6 Oct 2026.

Builds on

34 of 38 references.

Closest ancestors · 10

Same direction, newest first

  1. 2026SAFE-Pruner: Semantic Attention-Guided Future-Aware Token Pruning for Efficient Vision-Language-Action Manipulation28 May
  2. 2026See What Matters: Differentiable Grid Sample Pruning for Generalizable Vision-Language-Action Model12 May
  3. 2025Bridging the Semantic-Action Gap in Visual Token Pruning for Efficient VLA Inference20 Nov
  4. 2025Action-aware Dynamic Pruning for Efficient Vision-Language-Action Manipulation26 Sep
  5. 2025The Better You Learn, The Smarter You Prune: Towards Efficient Vision-language-action Models via Differentiable Token Pruning16 Sep
  6. 2025SpecPrune-VLA: Accelerating Vision-Language-Action Models via Action-Aware Self-Speculative Pruning6 Sep
  7. 2025SP-VLA: A Joint Model Scheduling and Token Pruning Approach for VLA Model Acceleration15 Jun
  8. 2025EfficientVLA: Training-Free Acceleration and Compression for Vision-Language-Action Models11 Jun
  9. 2025Think Twice, Act Once: Token-Aware Compression and Action Reuse for Efficient Inference in Vision-Language-Action Models27 May
  10. 2025VLA-Cache: Efficient Vision-Language-Action Manipulation via Adaptive Token Caching4 Feb

Other references · 24

  1. 2017Attention Is All You Needcited by 59,932
  2. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,084
  3. 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 12,865
  4. 2022Flow Matching for Generative Modelingcited by 5,426
  5. 2021RoFormer: Enhanced Transformer with Rotary Position Embeddingcited by 4,581
  6. 2022FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awarenesscited by 3,611
  7. 2023Sigmoid Loss for Language Image Pre-Trainingcited by 3,048
  8. 2024OpenVLA: An Open-Source Vision-Language-Action Modelcited by 2,777
  9. 2024π0: A Vision-Language-Action Flow Model for General Robot Controlcited by 2,297
  10. 2025π0.5: a Vision-Language-Action Model with Open-World Generalizationcited by 1,653
  11. 2023Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardwarecited by 1,490
  12. 2023LIBERO: Benchmarking Knowledge Transfer for Lifelong Robot Learningcited by 1,338
12 more
  1. 2025Fine-Tuning Vision-Language-Action Models: Optimizing Speed and Successcited by 844
  2. 2024PaliGemma: A versatile 3B VLM for transfercited by 679
  3. 2024An Image is Worth 1/2 Tokens After Layer 2: Plug-and-Play Inference Acceleration for Large Vision-Language Modelscited by 629
  4. 2024Scaling and evaluating sparse autoencoderscited by 530
  5. 2025SmolVLA: A Vision-Language-Action Model for Affordable and Efficient Roboticscited by 507
  6. 2024SparseVLM: Visual Token Sparsification for Efficient Vision-Language Model Inferencecited by 378
  7. 2024Prismatic VLMs: Investigating the Design Space of Visually-Conditioned Language Modelscited by 336
  8. 2024TinyVLA: Towards Fast, Data-Efficient Vision-Language-Action Models for Robotic Manipulationcited by 331
  9. 2024DeeR-VLA: Dynamic Inference of Multimodal Large Language Models for Efficient Robot Executioncited by 116
  10. 2025LapSum -- One Method to Differentiate Them All: Ranking, Sorting and Top-k Selectioncited by 12
  11. 2026Beyond Surrogate Gradients: Fully Differentiable Token Pruning for Vision-Language Modelscited by 7
  12. 2026The Gate, Not the Cache: Gate Provenance Bounds the Closed-Loop Reliability of Training-Free VLA Token Skippingcited by 2

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026BFA++: Hierarchical Best-Feature-Aware Token Prune for Multi-View Vision Language Action Model
  2. 2026DTP: A Simple yet Effective Distracting Token Pruning Framework for Vision-Language Action Models
  3. 2026Look Before Acting: Enhancing Vision Foundation Representations for Vision-Language-Action Models
  4. 2026AC^2-VLA: Action-Context-Aware Adaptive Computation in Vision-Language-Action Models for Efficient Robotic Manipulation
  5. 2025VITA-VLA: Efficiently Teaching Vision-Language Models to Act via Action Expert Distillation
  6. 2024CogACT: A Foundational Vision-Language-Action Model for Synergizing Cognition and Action in Robotic Manipulation
  7. 2026Learning to Accelerate Vision-Language-Action Models through Adaptive Visual Token Caching
  8. 2026Reducing Temporal Redundancy for Efficient Vision-Language-Action Inference