Q-VGM: 基于Q引导的值梯度匹配的流匹配VLA策略
Q-VGM: Q-Guided Value-Gradient Matching for Offline-to-Online RL of Flow-Matching VLA Policies
AI总结:
提出Q-VGM离线强化学习方法,通过将值梯度转化为去噪时间上的值梯度场,避免反向传播去噪链,高效微调流匹配VLA策略,在LIBERO等任务上显著提升成功率。
摘要:
我们提出Q引导的值梯度匹配(Q-VGM),一种离线强化学习方法,解决了微调流匹配视觉-语言-动作(VLA)策略中长期存在的挑战:如何高效地根据学习到的Q函数改进一个表达力强的流匹配动作专家。有效的改进必须利用评论家的一阶(梯度)信息,但这对于流策略很困难,因为直接通过其多步去噪过程反向传播值函数在VLA规模下数值不稳定,而策略梯度方法所需的可处理动作似然在迭代去噪下不可用。现有的基于值的方法要么通过整个去噪链反向传播,要么仅在测试时使用评论家而不更新策略,要么将评论家改进的动作作为终端标签蒸馏而不监督速度场。Q-VGM通过利用VGG-Flow(一种生成建模中流对齐的值梯度视角)绕过了这些问题,它将值梯度转化为去噪时间上的值梯度场,而不是不稳定的端到端目标。这不需要动作似然,也不需要反向传播去噪链,并且在一个固定的重放缓冲区上操作。评论家是一个动作敏感的Cal-QL集成,基于紧凑的RLT特征和每层动作注入。Q-VGM实现了一种实用的少样本初始化然后从经验中学习的范式:从少样本SFT pi0.5 VLA开始,该方法利用自生成的rollout数据显著提升任务性能,无需额外的专家监督。在LIBERO上,Q-VGM将平均成功率从75.0%提升到92.5%;在RoboTwin 2.0上,从76.4%提升到87.2%;在两个真实机器人桌面任务上,从40.0%提升到67.5%,在所有三种设置中均优于所有相同骨干、相同评论家的基线。
英文摘要:
We propose Q-Guided Value-Gradient Matching (Q-VGM), an offline-to-online RL method for fine-tuning flow-matching vision-language-action policies with a learned critic. Directly applying critic gradients to flow policies requires backpropagation through the multi-step denoising process (BPTT), which is costly and unstable at VLA scale. Motivated by an optimal-control view of denoising, we derive a local approximation connecting clean-action value gradients to local velocity corrections at late denoising steps. Q-VGM uses this connection to construct critic-guided targets for fine-tuning the pretrained action expert through velocity matching. Gradients pass only through local velocity predictions, avoiding backpropagation through the denoising trajectory. An action-sensitive chunk critic is trained with IQL offline and TD learning online. On LIBERO, Q-VGM improves the few-shot-SFT four-suite average from 84.6% to 90.7% offline using 150 policy rollouts per suite, and reaches 98.4% after online training. Compared with on-policy RL, Q-VGM achieves 5.3x higher sample efficiency on average across four LIBERO suites, measured by the episode budget for 95% rollout success rate. On three real-world bimanual tasks, offline Q-VGM improves average success from 66.7% to 98.3%.
机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。