轻量级机器人操作模型，能精准预测动作不确定性。
FabriVLA: A Lightweight Vision-Language-Action Model with Conformal Action Chunk Uncertainty
- 融合浅层与中间视觉语言层，保留细粒度图像特征。
- 在Meta-World上达90.0%成功率，仅0.88B参数。
- 新增校准方法可量化动作前缀风险，适合安全部署。
视觉-语言-动作（VLA）模型已成为通用机器人操作的主流范式，但其计算开销大且缺乏不确定性感知，限制了实际应用。本文提出FabriVLA，一种轻量级VLA模型，通过融合浅层与中间视觉语言模型（VLM）层，保留细粒度视觉特征；同时对动作标记的自注意力进行门控，使流匹配头仅在训练支持时引入步骤间结构。该模型端到端单阶段训练，仅需0.88B参数，在Meta-World MT50上达到90.0%平均成功率。我们进一步提出联合置信区间动作块校准（JCAC），一种后训练方法，为冻结策略添加轻量残差缩放头。单次策略查询即可将学习到的逐元素误差尺度转化为覆盖整个执行动作前缀的置信集，其均值半径比无条件置信集缩小3.3倍。在LIBERO-Safety上，该置信集可提前排序轨迹风险，支持风险优先审查。FabriVLA与JCAC共同构建了一个轻量、可审计且带校准动作不确定性的多任务操作框架。
原文摘要 · Abstract (English)
Vision-Language-Action (VLA) models have become a leading paradigm for general purpose robotic manipulation, but their computational cost and limited uncertainty awareness hinder practical deployment. We present FabriVLA, a lightweight VLA that fuses shallow and intermediate VLM layers to preserve fine-grained visual features, and gates self-attention among action tokens so that its flow matching head admits inter step structure only as far as training warrants. Trained end-to-end in a single stage, FabriVLA reaches a state-of-the-art 90.0\% average success on Meta-World MT50 with only 0.88B parameters. We further introduce Joint Conformal Action Chunk Calibration (JCAC), a post-training method that augments a frozen policy with a lightweight residual scale head. From a single policy query, JCAC turns a learned elementwise error scale into a set that covers the whole executed action prefix at a user chosen confidence level, 3.3$\times$ tighter in mean radius than an unconditional conformal set. On LIBERO-Safety, these bounds rank rollouts by risk before execution, supporting risk ranked review. Together, FabriVLA and JCAC provide a lightweight and auditable framework for multi-task manipulation with calibrated action uncertainty.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。