DIVA: Dual-Space Intent-Aware Visual Attenuation for Vision-Language-Action Policies
On the map: Vision-language-action models → VLA architectures and generalization also fits: Efficient VLA inference and adaptation
Abstract (arXiv)
Vision-language-action (VLA) policies typically feed dense visual patch tokens into a language-action backbone, preserving scene context but offering no explicit mechanism to regulate how strongly different visual tokens influence policy computation. We introduce DIVA, a Dual-Space Intent-Aware Visual Attenuation module with an anchor-then-attenuate design. DIVA combines high-level task intent with low-level visual evidence to estimate patch-wise relevance anchors, then applies them in two complementary spaces: it reweights projected visual tokens before backbone entry and persistently attenuates low-relevance visual states within the backbone. DIVA preserves the full visual token sequence and requires no external grounding supervision. On LIBERO, DIVA improves OpenVLA-OFT from 96.6% to 98.0% average success and raises its zero-shot LIBERO-Plus score from 69.6 to 72.6. Real-world experiments further show consistent gains under task-irrelevant visual perturbations, supporting the robustness of intent-aware visual attenuation beyond simulation.
Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in VLA architectures and generalization, one of 13 directions in Vision-language-action models. Builds on 22 of its 35 references that are in the map. First posted on arXiv 6 Oct 2026.
Builds on
22 of 35 references.
Closest ancestors · 1
Same direction, newest first
Other references · 21
- 2024π0: A Vision-Language-Action Flow Model for General Robot Controlcited by 2,407
- 2025Fine-Tuning Vision-Language-Action Models: Optimizing Speed and Successcited by 883
- 2024RDT-1B: a Diffusion Foundation Model for Bimanual Manipulationcited by 691
- 2025FAST: Efficient Action Tokenization for Vision-Language-Action Modelscited by 665
- 2021FILIP: Fine-grained Interactive Language-Image Pre-Trainingcited by 627
- 2025SpatialVLA: Exploring Spatial Representations for Visual-Language-Action Modelcited by 519
- 2023Vision-Language Foundation Models as Effective Robot Imitatorscited by 349
- 2024Robotic Control via Embodied Chain-of-Thought Reasoningcited by 346
- 2024TraceVLA: Visual Trace Prompting Enhances Spatial-Temporal Awareness for Generalist Robotic Policiescited by 304
- 2025LIBERO-Plus: In-depth Robustness Analysis of Vision-Language-Action Modelscited by 303
- 2025Interactive Post-Training for Vision-Language-Action Modelscited by 135
- 2025Unified Vision-Language-Action Modelcited by 134
9 more
- 2025NORA: A Small Open-Sourced Generalist Vision Language Action Model for Embodied Taskscited by 127
- 2025GraspVLA: a Grasping Foundation Model Pre-trained on Billion-scale Synthetic Action Datacited by 123
- 2025ReconVLA: Reconstructive Vision-Language-Action Model as Effective Robot Perceivercited by 81
- 2025The Better You Learn, The Smarter You Prune: Towards Efficient Vision-language-action Models via Differentiable Token Pruningcited by 36
- 2025Action-aware Dynamic Pruning for Efficient Vision-Language-Action Manipulationcited by 34
- 2025AVA-VLA: Improving Vision-Language-Action models with Active Visual Attentioncited by 25
- 2026Gaze-Regularized Vision-Language-Action Models for Robotic Manipulationcited by 9
- 2026DTP: A Simple yet Effective Distracting Token Pruning Framework for Vision-Language Action Modelscited by 8
- 2026ETA-VLA: Efficient Token Adaptation via Temporal Fusion and Intra-LLM Sparsification for Vision-Language-Action Modelscited by 1
Built on by
Not cited yet.
Similar, not cited
Close in wording, no citation either way
- 2025Think Twice, Act Once: Token-Aware Compression and Action Reuse for Efficient Inference in Vision-Language-Action Models
- 2026VisualThink-VLA: Visual Intermediate Reasoning for Effective and Low-Latency Vision-Language-Action Policies
- 2026Disentangling Spurious Correlations in Vision-Language-Action Models via Predicting Domain-Invariant Latent Lookahead
- 2026FiberTune: Preserving Action-Fiber Visual Residuals in Vision-Language-Action Fine-Tuning
- 2026One Token Per Frame: Reconsidering Visual Bandwidth in World Models for VLA Policy
- 2026Text-Vision Synergistic Token Caching: A Training-Free Framework for Efficient Vision-Language-Action Inference
- 2026Overcoming Visual Clutter in Vision Language Action Models via Concept-Gated Visual Distillation
- 2026The Layer Mystery of VLA: An Information-Theoretical Analysis of VLA Latent Interface