Overcoming Visual Clutter in Vision Language Action Models via Concept-Gated Visual Distillation

TL;DR

This paper introduces Concept-Gated Visual Distillation (CGVD), a novel inference framework that enhances vision-language action models' robustness in cluttered environments by suppressing distractors and stabilizing manipulation policies.

Contribution

The paper presents a training-free, model-agnostic method that improves VLA model performance in cluttered scenes through scene parsing, target refinement, and Fourier-based inpainting.

Findings

CGVD prevents performance collapse in cluttered environments.

Achieves 77.5% success rate in dense distractor scenarios, outperforming baseline.

Enforces strict attribute adherence for robust robotic manipulation.

Abstract

Vision-Language-Action (VLA) models demonstrate impressive zero-shot generalization but frequently suffer from a "Precision-Reasoning Gap" in cluttered environments. This failure is driven by background-induced feature dilution, where high-frequency semantic noise corrupts the geometric grounding required for precise manipulation. To bridge this gap, we propose Concept-Gated Visual Distillation (CGVD), a training-free, model-agnostic inference framework that stabilizes VLA policies. CGVD operates by parsing instructions into safe and distractor sets, utilizing a two-layer target refinement process--combining cross-validation and spatial disambiguation--to explicitly penalize false positives and isolate genuine manipulation targets. We then process the scene via Fourier-based inpainting, generating a clean observation that actively suppresses semantic distractors while preserving critical spatial geometry and visual proprioception. Extensive evaluations in highly cluttered manipulation tasks demonstrate that CGVD prevents performance collapse. In environments with dense semantic distractors, our method significantly outperforms state-of-the-art baselines, achieving a 77.5% success rate compared to the baseline's 43.0%. By enforcing strict attribute adherence, CGVD establishes inference-time visual distillation as a critical prerequisite for robust robotic manipulation in the clutter.