Guided Action Flow：Q 引导的流匹配 VLA 推理
Original title: Guided Action Flow: Q-Guided Inference for Flow-Matching Vision-Language-Action Policies
IndustryAI 66
Source: arXiv cs.ROPublish time unverified
arXiv:2607.02092v3 Announce Type: replace Abstract: Deploying a pretrained flow-matching vision-language-action (VLA) policy on a particular robot and workspace often calls for task-specific adaptation, while full- policy fine-tuning is costly and changes the base behavior. We present Guided Action Flow, an inference-time method that keeps a pretrained SmolVLA policy frozen and steers its reverse-time action-flow sampling with gradients from a task-specific action-chunk critic.