VLA-Rolling: Fewer Denoising Steps, Better Control via Adaptive Action Reuse

Key points

• We show that this discarded tail retains transferable denoising progress: action-tail initialization can match or outperform a native 10-step Gaussian baseline using fewer denoising steps.1
• Based on this observation, we introduce VLA-Rolling, a lightweight inference framework that reuses action tails through three stages: action-tail inheritance, per-action confidence estimation, and confidence-guided refinement through renoising and denoising budget allocation, while the underlying VLA remains frozen.1
• Across $π{0.5}$, SmolVLA, X-VLA, and MolmoAct2 on LIBERO, MetaWorld, LIBERO-Plus, and RoboTwin 2.0, VLA-Rolling reduces action-chunk inference cost in all 16 model--benchmark pairs, achieving up to a $1.48\times$ speedup, while improving success rate in 14 pairs by up to $15.08$ percentage points.1
• These results show that action tails are not merely discarded predictions, but reusable intermediate states that enable more efficient and accurate VLA inference.1

Sources (2)

• [1]VLA-Rolling: Fewer Denoising Steps, Better Control via Adaptive Action Reuse
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 18, 06:01 AM

  Flow-matching vision-language-action (VLA) models generate an action chunk by iteratively denoising Gaussian noise, but typically execute only a short prefix before replanning with updated observations.

  We show that this discarded tail retains transferable denoising progress: action-tail initialization can match or outperform a native 10-step Gaussian baseline using fewer denoising steps.

• [2]VLA-Rolling: Fewer Denoising Steps, Better Control via Adaptive Action Reuse
  Hugging Face Daily Papers · Oct 18, 12:00 AM · same content

Extractive summary: sentences quoted from the sources.