跳到正文
返回论文列表
cs.RO提交于 已译

SURE:面向高效视觉-语言-动作学习的时空不确定性表示

SURE:Spatiotemporal Uncertainty Representation for Efficient Vision-Language-Action Learning

Jingye Zhang · Choo Sin Wai · Dake Zhong · Zhifang Liu · Yuhong Zhang · Siyao Xiao · et al.

中文摘要

视觉-语言-动作(VLA)策略已成为端到端机器人控制的有前景范式,但其效果仍严重依赖大量示范数据。在操作任务中,这种依赖因动作空间中的结构化时空不确定性而加剧:多个邻近动作或轨迹都可能成功完成同一任务,而语义等价的示范常以不同速度展开。标准逐点监督将每个示范动作视为精确目标,在此类不确定性下造成不必要的学习负担。SURE 是一个数据高效的 VLA 训练框架,显式建模空间与时间不确定性。它在 VLA 主干上增加轻量转置卷积概率动作解码器,通过邻近动作 bin 间的局部相关性捕捉空间不确定性;并引入 Soft-DTW 对齐监督,以适应预测动作块与示范动作块之间的局部时序变化。两个组件共同降低对示范者特定姿态与执行时序的敏感性,实现更数据高效的动作学习。在 LIBERO、RoboTwin 2.0 和真实世界操作任务上评估,仅用 10% 示范数据或 20% 训练步数,SURE 的操作成功率即匹配或超过基线 VLA。

关键要点

  1. 01问题:VLA 策略依赖大量示范数据,动作空间存在多解与速度差异带来的时空不确定性。
  2. 02方法:轻量转置卷积概率动作解码器建模空间不确定性,Soft-DTW 对齐监督处理时序变化。
  3. 03结果:仅用 10% 示范数据或 20% 训练步数,成功率匹配或超过基线 VLA。
  4. 04验证:在 LIBERO、RoboTwin 2.0 及真实世界操作任务上评估。

解读

尚无解读。

原始英文摘要

arXiv:2611.09684v1 Announce Type: new Abstract: Vision-language-action (VLA) policies have become a promising paradigm for end-to-end robot control, but their effectiveness still depends heavily on large volumes of demonstration data. In manipulation tasks, this dependence is compounded by structured spatiotemporal uncertainty in the action space: multiple nearby actions or trajectories may successfully solve the same task, while semantically equivalent demonstrations often unfold at different speeds. Standard pointwise supervision treats each demonstrated action as an exact target, imposing an unnecessary learning burden under such uncertainty. We present SURE, a data-efficient VLA training framework that explicitly models both spatial and temporal uncertainty. SURE augments a VLA backbone with a lightweight transposed-convolutional probabilistic action decoder that captures spatial uncertainty through local correlations among neighboring action bins. It further introduces Soft-DTW alignment supervision to accommodate local timing variations between predicted and demonstrated action chunks. Together, these two components reduce sensitivity to demonstrator-specific poses and execution timing, enabling more data-efficient action learning. We evaluate SURE on LIBERO, RoboTwin 2.0, and real-world manipulation tasks. Even with only 10% of the demonstration data or 20% of the training steps, SURE achieves manipulation success rates that match or surpass baseline VLAs.

同方向论文 · cs.RO

查看全部 →

反向发现 · 这篇论文相关的工具与提示词

去交叉发现 →

关联工具 · 1

Google DeepMind78

Google DeepMind 发布 Gemini Robotics 2.5 视觉-语言-动作模型

Google DeepMind 发布 Gemini Robotics 2.5,称其为目前能力最强的视觉-语言-动作(VLA)模型,在200项任务上相比此前基线成功率提升30%至100%,完成任务速度快20%至30%。

推荐理由:原文给出200项任务评测、Mercedes-Benz产线96%成功率与安全指标,可判断VLA模型在真实场景的进展。