利用可行动作邻域先验提升视觉-语言-动作模型微调
Boosting Vision-Language-Action Finetuning with Feasible Action Neighborhood Prior
对做 VLA/机器人策略学习的人，这篇给出了一个可直接复用的正则化手段：把「每个状态存在一族近似等价动作」这一物理先验编码成高斯输出分布约束，替代从语言模型直接继承的交叉熵式训练，从而在 SFT 与 RFT 两种微调范式下同时提升样本效率与 OOD 成功率。
原文依据 we introduce a Gaussian prior that promotes locally smooth and unimodal predictions around the preferred direction and magnitude
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论在真实世界的机器人操作中，状态通常对应着一族近似等价的动作。也就是说，对每个状态而言，存在的是一个可行动作邻域（feasible action neighborhood, FAN），而非单一的正确动作；在该邻域内，不同运动带来的进展无法区分。然而，当前主流的 VLA 训练方法直接沿袭自语言建模的设定，并未利用 FAN 这一性质，因而导致泛化性差、样本效率低。为弥补这一局限，我们引入一个 FAN 引导的正则化器，将模型的输出分布塑造成与 FAN 的几何结构相一致。具体而言，我们引入一个高斯先验，促使模型在偏好的方向与幅值附近产生局部平滑且单峰的预测。在强化微调（RFT）与监督微调（SFT）两种设定下的大量实验中，我们的方法在样本效率以及分布内与分布外（OOD）场景下的成功率上均取得显著提升。通过与物理操作内在的动作容差对齐，FAN 引导的正则化为样本高效且可泛化的 VLA 适配提供了一种有原理依据且实用的方法。源代码已公开于：https://github.com/Haochen-Niu/FAN-VLA。
英文原文摘要 1275 字符
In real-world robotic manipulation, states typically admit a neighborhood of near-equivalent actions. That is, for each state, there exist a feasible action neighborhood (FAN) rather than a single correct action, within which motions yield indistinguishable progress. However, prevalent VLA training methodologies are directly inherited from linguistic settings and do not exploit the FAN property, thus leading to poor generalization and low sample efficiency. To address this limitation, we introduce a FAN-guided regularizer that shapes the model's output distribution to align with the geometry of FAN. Concretely, we introduce a Gaussian prior that promotes locally smooth and unimodal predictions around the preferred direction and magnitude. In extensive experiments across both reinforced finetuning (RFT) and supervised finetuning (SFT), our method achieves significant improvement in sample efficiency, and success rate in both in-distribution and out-of-distribution (OOD) scenarios. By aligning with the intrinsic action tolerance of physical manipulation, FAN-guided regularization provides a principled and practical method for sample-efficient, and generalizable VLA adaptation. The source code is publicly available at: https://github.com/Haochen-Niu/FAN-VLA.