AI Paper Atlas

Research map / Vision-language and multimodal / Robot action and navigation

$\text{A}^3$-VLA: Automatic Perception-Guided Attention Alignment for Robot Manipulation

Shengzhe Zhang, Qi Zhang, Dazhong Shen, Chao Wang, Shaopeng Zhai, Hui Xiong

NeurIPS 2026 · Vision-language and multimodal › Robot action and navigation · approach: Vision-language-action foundation models

Read the paper

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes A^3-VLA, which aligns a VLA's attention with task-relevant regions (target objects and interaction areas) using automatically generated perception guidance, to counter reliance on task-irrelevant visual shortcuts.

Problem

Vision-language-action models trained only with action prediction lack fine-grained task-specific perceptual supervision, so they rely on task-irrelevant visual shortcuts and attend away from target objects and key interaction areas.

Approach

Automatic perception-guided attention alignment: supervises the VLA's attention toward target objects and interaction regions using automatically derived perceptual signals, rather than relying on action loss alone. Based on the abstract only; details unknown.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add language-to-image attention allocation loss (ablation)A3-VLA (pi0) · VLABench (5 suites avg) ablationsuccess rate: 0.52 -> 0.64Improved
add state-to-image attention allocation loss (ablation)A3-VLA (pi0) · VLABench (5 suites avg) ablationsuccess rate: 0.60 -> 0.64Improved
add contrastive language-target loss (ablation)A3-VLA (pi0) · VLABench (5 suites avg) ablationsuccess rate: 0.61 -> 0.64Improved
add A3-VLA guidance to pi0pi0 · VLABench (5 suites avg)success rate: 0.47 -> 0.64Improved
add A3-VLA guidance to OpenVLA-OFTOpenVLA-OFT · VLABench (5 suites avg)success rate: 0.16 -> 0.40Improved
add A3-VLA guidance to pi0.5pi0.5 · VLABench (5 suites avg)success rate: 0.41 -> 0.67Improved
add A3-VLA guidancepi0 / OpenVLA-OFT / pi0.5 · RoboTwin 2.0 (4 tasks avg)success rate: 0.42 -> 0.52 (pi0)Improved
add A3-VLA guidance to pi0pi0 · Real-world select bottlessuccess rate: 0.25 -> 0.75Improved
add A3-VLA guidance under texture perturbationpi0.5 · VLABench TPsuccess rate: 0.18 -> 0.42Improved
add A3-VLA guidance under Gaussian noisepi0.5 · VLABench NPsuccess rate: 0.33 -> 0.57Improved
increase attention-loss weight lambda1 or contrastive weight lambda2 (ablation)A3-VLA (pi0) · VLABenchsuccess rateHurt
raise attention threshold eta1 beyond moderate value (ablation)A3-VLA (pi0) · VLABenchsuccess rateHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy