VLA-Rolling: Fewer Denoising Steps, Better Control via Adaptive Action Reuse
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. Under this receding-horizon paradigm, the unexecuted remainder---the action tail---is discarded, and the next chunk is regenerated from fresh noise. We show that this discarded
Reading in Plain English mode — toggle below to switch
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. Under this receding-horizon paradigm, the unexecuted remainder---the action tail---is discarded, and the next chunk is regenerated from fresh noise. We show that this discarded
This research addresses a critical efficiency challenge in VLA models for robotic control, where significant computational resources are wasted on discarded action tails. By proposing adaptive action reuse, it offers a method to achieve better control with fewer denoising steps, directly impacting the real-time performance and operational costs of AI agents.