让机器人在不同摄像头视角下仍能准确执行任务，提升视觉-语言-动作策略的鲁棒性。
Cross-View Action Consistency for Camera-Robust Vision-Language-Action Policies
- 通过构建动作等价的视图对，强制模型关注场景本身而非摄像头位置。
- 在真实机器人上，跨摄像头任务成功率从53.3%提升至74.4%。
- 无需深度或相机参数，适合实际部署中的多视角应用。
从固定摄像头场景微调的视觉-语言-动作（VLA）策略在摄像头移动时会失效，即使任务、物体、语言和机器人状态不变。本文仅使用场景RGB图像、语言和本体感知信息，不依赖摄像头标签、外参、深度或点云输入，研究场景视角鲁棒性。为防止手腕视觉流成为未扰动的捷径，全程屏蔽腕部流。针对基于光流的VLA，提出正则化动作光流速度场——直接积分生成连续动作块的核心量。通过重置原始LIBERO演示至同一MuJoCo状态，并渲染名义与扰动摄像头视角，构建动作等价视图对。双视图均通过光流匹配监督，同时引入跨视图损失，使相同采样坐标处预测的动作光流速度一致。在LIBERO-Plus摄像头扰动赛道上，本方法达87.2±0.4%（每种子3次训练共4,797次回放），较仅用光流匹配提升7.4个百分点（79.8±0.8%，同数据集3种子），较朴素混合摄像头微调提升12.5个百分点；保持名义摄像头性能95.0±0.8%（同数据集光流匹配：95.0±4.3%）。随机配对对照组下降至25.8%，证明增益依赖于动作等价配对。在真实机器人上，3个桌面任务各10次回放，跨摄像头成功率从53.3%提升至74.4%，使用相同的单场景RGB推理接口。
原文摘要 · Abstract (English)
Vision-language-action (VLA) policies fine-tuned from a fixed scene camera can fail when the camera is moved, even when the task, objects, language, and robot state are unchanged. We study scene-camera viewpoint robustness using only a scene RGB image, language, and proprioception, without camera labels, extrinsics, depth, or point-cloud inputs. The wrist stream is masked throughout to prevent an unperturbed visual shortcut from confounding attribution to scene-camera variation. For flow-based VLAs, we propose to regularize the action-flow velocity field, the quantity directly integrated to generate continuous action chunks. We construct action-equivalent view pairs by resetting original LIBERO demonstrations to the same MuJoCo state and rendering nominal and perturbed scene-camera views. Both views are supervised by flow matching, while a cross-view loss encourages their predicted action-flow velocities to agree at the same sampled flow coordinates. On the LIBERO-Plus camera-perturbation track, our method reaches 87.2$\pm$0.4% (4,797 rollouts per seed across 3 training seeds), +7.4pp over flow-matching-only training on the same paired data (79.8$\pm$0.8%, also 3 seeds) and +12.5pp over naive mixed-camera SFT, while maintaining nominal-camera ID performance (95.0$\pm$0.8%; same-data FM-only: 95.0$\pm$4.3%). A shuffled-pair control collapses to 25.8%, showing that the gain depends on action-equivalent pairing. On a real robot, we evaluate three tabletop tasks with 10 rollouts per task and camera placement; held-out-camera success improves from 53.3% to 74.4% under the same single-scene-RGB inference interface.
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。