数据工程与质量 · Data Engineering and Quality

近一年具身智能论文中的数据时空一致性

Spatiotemporal Consistency in Embodied AI Data

更新于

30 秒结论

先读这个判断

本文考察 2025 年 7 月 27 日至 2026 年 7 月 27 日的具身智能论文,问题不是“哪些方法显式使用了 spatiotemporal consistency 这个词”,而是:论文在什么位置讨论数据的时间、空间、动作与物理后果必须彼此一致。检索池包含 1,401 篇候选论文,52 篇具有完整可读的非 OCR 全文;本文从中接纳 20 篇完成全文阅读与论据核验的论文,覆盖数据采集、多模态感知、4D 世界模型、第一视角迁移和 VLA 部署。

精读论文
20
正式证据事件
20
知识单元
EA-DATA, EA-SENSOR, EA-4D, EA-ALIGN
审计状态
当前结算 run 已通过仓库校验

v0 · 完整正文

研究综述

研究边界

本文考察 2025 年 7 月 27 日至 2026 年 7 月 27 日的具身智能论文,问题不是“哪些方法显式使用了 spatiotemporal consistency 这个词”,而是:论文在什么位置讨论数据的时间、空间、动作与物理后果必须彼此一致。检索池包含 1,401 篇候选论文,52 篇具有完整可读的非 OCR 全文;本文从中接纳 20 篇完成全文阅读与论据核验的论文,覆盖数据采集、多模态感知、4D 世界模型、第一视角迁移和 VLA 部署。

这一范围可以支持机制归纳,但不能给出跨任务统一的同步误差、帧率或标定精度阈值。多数论文报告的是特定机器人、传感器和任务下的实现与消融;例如 10 Hz、6 FPS 或快慢流频率比都是系统实例,不是行业标准。

中心判断

近一年的变化不是出现了一个新的“时空一致性分数”,而是研究共同把一致性从采集端的一次性检查,推进为贯穿数据、模型与闭环执行的五层契约:共同物理时间、共同空间参考、跨视角/跨模态对应、动作—状态因果匹配,以及长序列中的对象身份与接触连续性。时间戳相同只能证明记录层对齐;只有当同一动作在同一坐标语义下对应同一段物理状态变化,并在预测和真实执行中保持同步,数据才对策略学习真正一致。

这是本文的跨论文推断。它会被以下结果证伪:只做时间戳与外参对齐、无需动作语义、接触事件和长程后果检核的数据,在跨本体与真实闭环中仍能稳定达到同等性能。现有证据恰好指向相反方向:异构数据若不同时处理空间、形态、物理时间和标签可靠性,会损害动作学习;离线指标也可能在数据—模型—控制接口不一致时无法转化为稳定真机行为。ACE-Ego-0 与一项 UR5 实机研究 分别从预训练消融和部署观察给出了这一边界。

一、时间一致性:统一的不是帧号,而是物理事件

最底层的要求仍然是硬件同步。HapTile 在机器人控制环内同步视觉、触觉、语言、状态和动作,并检查时间戳缺口、损坏轨迹与动作—状态一致性;Visuo-Tactile World Models 则先用时间戳对齐腕部位姿、关节、外部视觉和双指触觉视频,再降采样训练。这些工作说明,模型输入频率可以降,但原始物理时间与对齐过程不能丢。HapTile 的价值也不只是“多一种触觉数据”,而是把多个模态放回同一控制事件中。

更重要的变化是多速率建模。视觉和语言承担较慢的场景理解与任务约束,触觉在接触后进入更快的纠偏回路;AT-VLA 因此把慢速视觉—语言流和快速触觉流分开。H-WM 也用低频逻辑状态维持长任务顺序,用视觉子目标连接感知,再让高频 VLA 执行动作块。这里的“一致”不再等于所有通道采相同 FPS,而是不同频率的信号能被映射到同一段物理过程:接近、接触、滑移、释放和恢复。

因此,按固定帧数切 action chunk 是危险的。不同来源即使都提供 16 帧,实际覆盖的物理时长、控制延迟和接触阶段也可能不同。Embodied Image Compression 的 10 Hz 多视角管线只是一个 UR5 实例;合理的数据契约应保存实际采样时刻、时钟源、丢帧与抖动,再按物理时长或事件边界生成训练窗口。

二、空间一致性:共同坐标比更多视角更优先

空间一致性首先是参考系问题。第一视角视频中的相机与手腕同时运动;ActiveMimic 明确指出,直接把腕部轨迹当动作监督,会把相机平移和旋转混入手部运动。类似问题也出现在跨本体预训练:同一个数值向量可能分别表示相机坐标下位移、末端绝对位姿或关节命令。没有坐标图、单位、动作语义和形态条件,“数值接近”并不代表物理动作一致。

第二层是跨视角几何对应。MVISTA-4D 从单视角 RGBD 生成几何一致的任意视角,再把它们反投影、融合成随时间变化的 3D 结构;Embody4D 则把固定或稀疏机位导致的局部观测,转化为新视角视频与 3D 感知的数据扩增问题。这两条路线扩展了观测面,但也隐含一个质量闸门:新视角必须保持对象身份、相机几何和时间连续,不能仅靠单帧视觉逼真验收。

第三层是可见性与标定不确定性。3PoinTr 不因关键点短暂遮挡就删除整条轨迹,而是只屏蔽不可见的点—时刻损失,从而保留跨帧对象运动监督。RGB-S 将触觉接触投影到图像域时,还用空间分布显式表达运动学和标定误差。它们共同说明:空间对齐不应伪装成无误差真值,而要同时记录对应关系、可见性和残差。

三、跨模态一致性:同一时刻不等于同一物理含义

接触任务最能暴露“同时间戳帧”的不足。相机可能仍看到夹爪覆盖物体,却无法判断摩擦、滑移和局部形变;触觉或力信号虽然时间上对齐,也可能因安装位置和表示空间不同而无法直接融合。ContactWorld 的比较显示,稳定长程规划依赖空间结构、时间连续性和跨模态兼容性同时成立;在其 12 个接触任务上,点云与触觉力场的组合优于单纯腕部或前视图,但论文没有把“增加模态”本身当作充分条件。

Dream-Tac 把动作块、未来视觉和未来触觉置于同一预测目标中,实际上把跨模态对齐改写为“动作之后,各模态应如何共同变化”。HT-Bench 则把规模推进到约 1,000 万 RGB 帧、780 万触觉帧和 226 项任务,并评估接触结构、跨模态对应和时间动态。大规模配对数据使表征研究成为可能,但仍不能自动外推为真实机器人闭环收益。

多模态带宽也应服从动作效用。SPARC 指出,多机位数据不应把码率均匀分给每个视图和区域,而应保留对当前动作有用的内容。Event-VLA 以事件流补充光照变化下退化的 RGB,进一步说明模态价值来自它是否补足特定不可观测状态,而非传感器数量。

四、因果一致性:动作必须与状态变化成对

真正决定数据能否训练控制策略的,是动作与后果是否匹配。Kinema4D 把机器人控制表示为运动学正确的 4D 轨迹,再让生成模型预测环境响应;这比仅以低维动作向量条件化视频更明确地绑定了“机器人如何运动”和“世界如何变化”。HapTile 的动作—状态检核、Dream-Tac 的动作—视觉—触觉联合预测,都是同一方向:数据单元应是一次干预及其后果,而非彼此松散的帧和标签。

这一层也给出了最强反例。BadWAM 表明,世界—动作模型可以“想象得合理却执行得错误”;安全属性不应只检查生成未来是否逼真,而要验证想象未来与实际将执行动作是否同步。这使“动作—想象一致性”成为世界模型数据和评测的新要求。若预测视频与真实动作解绑,再好的时空画面也不能作为可靠控制证据。

五、长程一致性:要保持身份、接触和任务进度

长序列误差并不只来自画面漂移。对象身份交换、遮挡后重现错误、接触关系中断、动作延迟累积和任务阶段错位,都会让局部合理的预测在几秒后失去物理意义。WEAVER 将有效机器人世界模型概括为结果保真、长程时间一致和计算效率三者同时满足;缺少其中任何一项,都难以用于策略评估、改进或规划。

这也解释了为什么 H-WM 的多时间尺度结构有意义:低频逻辑维持全局任务顺序,高频控制吸收局部变化。但该路线仍有边界。稀疏子目标可能漏掉接触瞬态,长程一致的视觉计划也可能与真实执行动作不同步。因而部署验收必须同时检查任务进度、对象/接触连续性和动作忠实,而不能只看终帧或视频质量。

可操作的数据契约

层级入库必须保留建议检核
物理时间原始时间戳、时钟源、采样率、丢帧/抖动、控制延迟跨模态时间残差、事件顺序、物理窗口长度
空间参考坐标图、单位、内外参、标定版本、机位/硬件 ID重投影残差、参考系闭环、遮挡与可见性
模态对应视觉、深度、触觉、力、本体状态的字段白名单与缺失标记对应完整率、模态污染消融、跨实例退化
动作因果动作语义、控制频率、执行延迟、动作前后状态、接触事件动作—状态残差、接触因果顺序、不可达率
长程动态对象/点轨迹、阶段、失败、接管、恢复、奖励或进度身份保持、长程漂移、动作—想象同步、闭环成功率

这张表的含义不是强迫所有数据源提供全部字段。人类视频、机器人示教、触觉序列和合成 4D 数据可以各自缺字段,但缺失必须显式标注,训练时按可靠监督字段使用。异构可以扩大内容覆盖,一致性则应集中在数据契约和目标机器人锚点上。

条件、分歧与研究空白

第一,目前没有证据支持统一帧率、码率或同步误差阈值。任务时域、控制频率、传感器延迟和接触速度不同,阈值必须通过目标策略的闭环曲线标定。

第二,生成式新视角和 4D 伪标注能扩大数据覆盖,但仍缺少跨方法统一的几何—动作—接触联合验收。视觉一致不能替代物理一致,表征分数也不能替代闭环控制。

第三,多速率系统缺少统一的端到端延迟预算。慢速语义、视觉子目标、动作块和快速触觉纠偏如何在同一时间轴上对账,仍是工程与研究共同空白。

第四,硬件长期漂移、换件和重标定尚未被多数数据集作为核心变量。空间一致性若只在采集首日成立,就不能支撑长期部署。

结论

近一年论文给出的最重要变化,是把数据时空一致性从“传感器同步是否成功”推进到“数据能否忠实描述一次物理干预及其后果”。可靠的数据不必同构,也不必同频;但它必须回答五个问题:何时发生、在什么坐标中发生、哪些观测对应同一状态、哪个动作导致了什么变化,以及这种关系能否跨遮挡、接触和长任务保持。对具身数据工程而言,下一步不应再寻找一个全局一致性分数,而应建立可分层审计、可在真实闭环中证伪的数据契约。

References

Accepted evidence

去重论文引用

  1. $\texttt{WEAVER}$, Better, Faster, Longer: An Effective World Model for Robotic ManipulationArnav Kumar Jain, Yilin Wu, Jesse Farebrother, Gokul Swamy, Andrea Bajcsy · 2026
  2. 3PoinTr: 3D Point Tracks for Learning Manipulation from Unconstrained Human VideosAdam Hung, Bardienus Pieter Duisterhof, Jeffrey Ichnowski · 2026
  3. ACE-Ego-0: Unifying Egocentric Human and Robotic Data for VLA PretrainingHao Li, Ganlong Zhao, Yufei Liu, Haotian Hou, Guoquan Ye, Tongyan Fang, Chunxiao Liu, Siyuan Huang et al. · 2026
  4. ActiveMimic: Egocentric Video Pretraining with Active PerceptionXingyao Lin, Guojin Zhong, Tianyi Lu, Ziyi Ye, Yichen Zhu, Zuxuan Wu, Yu-Gang Jiang · 2026
  5. AT-VLA: Adaptive Tactile Injection for Enhanced Feedback Reaction in Vision-Language-Action ModelsXiaoqi Li, Muhe Cai, Jiadong Xu, Juan Zhu, Hongwei Fan, Yan Shen, Guangrui Ren, Hao Dong · 2026
  6. BadWAM: When World-Action Models Dream Right but Act WrongQi Li, Xingyi Yang, Xinchao Wang · 2026
  7. ContactWorld: What Matters in Vision-Tactile World Models for Contact-Rich ManipulationZhiyuan Zhang, Pokuang Zhou, Kaidi Zhang, Adeesh Desai, Temitope Amosa, Davood Soleymanzadeh, Jiuzhou Lei, Minghui Zheng et al. · 2026
  8. Dream-Tac: A Unified Tactile World Action Model for Contact-Rich Robot ManipulationYunfan Lou, Yifan Ye, Yankai Fu, Jun Cen, Xiaowei Chi, Yaoxu Lyu, Peidong Jia, Sirui Han et al. · 2026
  9. Embodied Image Compression: Towards Codec for Robotic Visual SystemsZhenghao Chen, Zijie Yue, Haozhe Li, Yifan Zhao, Yunhui Guo, Zhi Wang · 2025
  10. Embody4D: A Generalist Data Engine for Embodied 4D World ModelingPeiyan Tu, Hanxin Zhu, Jingwen Sun, Shaojie Ren, Cong Wang, Yuyan Xu, Jiayi Luo, Xiaoqian Cheng et al. · 2026
  11. Event-VLA: Action-Conditioned Event Fusion for Robust Vision-Language-Action ModelJiaxin Liu, Xun Xu, Zhenhao Zhang, Hanqing Wang, Ruiqi Chen, Shi Chang, Weiyu Guo, Laurent Kneip · 2026
  12. H-WM: Robotic Task and Motion Planning Guided by Hierarchical World ModelJinbang Huang, Wenyuan Chen, Zhiyuan Li, Oscar Pang, Xiao Hu, Lingfeng Zhang, Yuanzhao Hu, Zhanguang Zhang et al. · 2026
  13. HapTile: A Haptic-Informed Vision-Tactile-Language-Action Dataset for Contact-Rich Imitation LearningAmirhosein Alian, Yongqiang Zhao, Shiyi Gu, Xuyang Zhang, Zhuo Chen, Christopher E. Mower, Haitham Bou-Ammar, Shan Luo · 2026
  14. HT-Bench: Benchmarking and Learning Dexterous Full-Hand Tactile Representations with Egocentric VisionYuzhe Huang, Jiaping Wu, Jiaming Jiang, Hezhe Lin, Aikebaier Aierken, Yunlong Wang, Kun Cheng, Ziyuan Jiao et al. · 2026
  15. Kinema4D: Kinematic 4D World Modeling for Spatiotemporal Embodied SimulationMutian Xu, Tianbao Zhang, Tianqi Liu, Zhaoxi Chen, Xiaoguang Han, Ziwei Liu · 2026
  16. MVISTA-4D: View-Consistent 4D World Model with Test-Time Action Inference for Robotic ManipulationJiaxu Wang, Yicheng Jiang, Tianlun He, Jingkai Sun, Qiang Zhang, Junhao He, Jiahang Cao, Zesen Gan et al. · 2026
  17. RGB-S: Image-Aligned Tactile Saliency for Robust Dexterous ManipulationShengcheng Luo · 2026
  18. SPARC: Spatially Adaptive Rate Control for Vision-Language-Action ModelsSangYun Chung, Mincheol Shin, Jihyun Kim, Sangpil Kim · 2026
  19. Vision-Language-Action Models: Experimental Insights from a Real-World UR5 PlatformMathilde Hochedel, Marc Lalonde · 2026
  20. Visuo-Tactile World ModelsCarolina Higuera, Sergio Arnaud, Byron Boots, Mustafa Mukadam, Francois Robert Hogan, Franziska Meier · 2026
完整证据附录(20 条正式证据事件)
  • Time range: 2025-07-27..2026-07-27
  • Events: 20
  • 每个事件一节,标题即锚点;trace-map 中的 event 链接跳转到这里。

EA-4D-READ-0014

  • Claim: MVISTA-4D formulates embodied 4D prediction as view-consistent arbitrary-view RGBD generation from a single-view RGBD observation and fuses the generated views into a more complete 3D structure over time.
  • Stance: support | Confidence: direct
  • Paper: 2602.09878 MVISTA-4D: View-Consistent 4D World Model with Test-Time Action Inference for Robotic Manipulation
  • Locator: Abstract (full-text section)
  • Evidence: The abstract describes single-view RGBD input, arbitrary-view RGBD generation, and back-projection/fusion as the route to complete time-varying 3D structure.
  • Quote: “Abstract World-model-based imagine-then-act becomes a promising paradigm for robotic manipulation, yet existing approaches typically support either purely image-based forecasting or reasoning over partial 3D geometry, limiting their ability to predict complete 4D scene dynamics. To solve this, this work explores a novel embodied 4D world model that enables geometrically consistent, arbitrary-view RGBD generation: given only a single-view RGBD observation as input, the model “imagines” the remain”
  • Authors: jiaxu-wang; yicheng-jiang; tianlun-he; et al.

EA-4D-READ-0005

  • Claim: Kinema4D argues that robot-world interaction should be simulated as a 4D event: robot control is represented with kinematically correct 4D trajectories, while a generative model predicts environment reactions.
  • Stance: support | Confidence: direct
  • Paper: 2603.16669 Kinema4D: Kinematic 4D World Modeling for Spatiotemporal Embodied Simulation
  • Locator: Abstract (full-text section)
  • Evidence: The method disentangles precise robot control from generative environmental reaction by driving a URDF robot through kinematics, projecting a 4D robot pointmap sequence, and jointly generating synchronized RGB/pointmap futures.
  • Quote: “Abstract Simulating robot-world interactions is a cornerstone of Embodied AI. Recently, a few works have shown promise in leveraging video generations to transcend the rigid visual/physical constraints of traditional simulators. However, they primarily operate in 2D space or are guided by static environmental cues, ignoring the fundamental reality that robot-world interactions are inherently 4D spatiotemporal events that require precise interactive modeling. To restore this 4D essence while ensu”
  • Authors: mutian-xu; tianbao-zhang; tianqi-liu; et al.

EA-4D-READ-0015

  • Claim: Embody4D targets the sparse-view limitation of robot video data with monocular-to-novel-view video transformation and a 3D-aware compositional synthesis pipeline for training data.
  • Stance: support | Confidence: direct
  • Paper: 2605.01799 Embody4D: A Generalist Data Engine for Embodied 4D World Modeling
  • Locator: Abstract (full-text section)
  • Evidence: The abstract ties fixed or sparse viewpoints to partial observations and introduces both novel-view video generation and a compositional synthesis pipeline to address data scarcity.
  • Quote: “Abstract Embodied agents require robust and comprehensive 3D spatiotemporal representations to support spatial reasoning, manipulation understanding, and downstream decision making. However, existing robot data are typically captured from fixed or sparse viewpoints, providing only partial and view-dependent observations, which limits multi-view perception and generalization across viewpoints. Given the difficulty of collecting additional viewpoints in real-world settings, we propose Embody4D, a”
  • Authors: peiyan-tu; hanxin-zhu; jingwen-sun; et al.

EA-4D-READ-0013

  • Claim: Dream-Tac将动作块、未来视觉和未来触觉放进同一世界动作建模目标,以弥补单纯视觉未来对接触互动线索的不足。
  • Stance: support | Confidence: direct
  • Paper: 2606.08737 Dream-Tac: A Unified Tactile World Action Model for Contact-Rich Robot Manipulation
  • Locator: 3.1. Problem Formulation
  • Evidence: 问题建模段先定义动作与视觉未来的联合分布,再明确把未来触觉纳入联合预测目标。
  • Quote: “Building on these two formulations, a world action model combines action prediction and future observation prediction into a unified framework. Specifically, it jointly models (3) or equivalently factorizes the joint distribution as (4) where future visual prediction provides predictive structure for action generation. However, in contact-rich manipulation, vision alone is often insufficient to capture physical interaction cues. To address this limitation, we introduce Dream-Tac, an enhanced wor”
  • Authors: yunfan-lou; yifan-ye; yankai-fu; et al.

EA-4D-READ-0003

  • Claim: WEAVER defines useful robot world models by three joint requirements: fidelity to real outcomes, temporal consistency over long horizons, and enough efficiency for evaluation, improvement, and planning.
  • Stance: support | Confidence: direct
  • Paper: 2606.13672 $\texttt{WEAVER}$, Better, Faster, Longer: An Effective World Model for Robotic Manipulation
  • Locator: 3 WEAVER : World Estimation Across Views for Embodied Reasoning
  • Evidence: The paper argues that manipulation world models must satisfy fidelity, consistency, and efficiency together, then designs a multi-view latent world model with reward/value prediction to support policy evaluation, synthetic policy improvement, and test-time planning.
  • Quote: “Figure 2 : WEAVER Architecture. Left: The world model encodes memory, history, and action sequences to image future rollouts in latent space. Middle: The latent verifier, equipped with reward and critic heads, selects samples with high advantage to steer the policy distribution. Right: Decoded generation corresponding to different outcomes of action sequences. We now describe the key ingredients in WEAVER : a robot world model designed to support policy evaluation, policy improvement, and test-t”
  • Authors: arnav-kumar-jain; yilin-wu; jesse-farebrother; et al.

EA-TWM-READ-0003

  • Claim: 在接触丰富操作中,触觉世界模型的关键不是简单增加模态,而是让表征同时具备空间结构、时间连续性和跨模态兼容性。
  • Stance: support | Confidence: direct
  • Paper: 2606.13877 ContactWorld: What Matters in Vision-Tactile World Models for Contact-Rich Manipulation
  • Locator: Abstract (full-text section)
  • Evidence: ContactWorld 在 12 个接触丰富任务上比较视觉与触觉表征;点云把平均规划成功率从腕部视角 20.7% 和前视 22.0% 提升到 32.1%,点云加触觉力场进一步到 36.1%。作者强调触觉效果取决于跨模态表征兼容,而非模态数量本身。
  • Quote: “Abstract Contact-rich manipulation requires world models to reason over complex contact dynamics from multimodal sensory observations. However, it remains unclear which representation properties fundamentally support stable long-horizon planning in contact-rich settings. In this paper, we present ContactWorld, a benchmark and systematic empirical study of vision-tactile world models spanning 12 contact-rich manipulation tasks, including insertion, disassembly, screwing, and exploratory interacti”
  • Authors: zhiyuan-zhang; pokuang-zhou; kaidi-zhang; et al.

EA-PRETRAIN-DATA-2026-0003

  • Claim: 多相机 VLA 不应把码率在机位和画面区域间均分;应优先保留对当前动作有用的视图和区域。
  • Stance: support | Confidence: direct
  • Paper: 2606.16253 SPARC: Spatially Adaptive Rate Control for Vision-Language-Action Models
  • Locator: 1 Introduction and 3 Method
  • Evidence: 论文指出不同机位和图像区域对控制的价值不均匀,SPARC 通过时序 mask 自适应分配比特。
  • Quote: “Uniform bitrate allocation across cameras and image regions is therefore fundamentally inefficient.”
  • Authors: sangyun-chung; mincheol-shin; jihyun-kim; et al.

EA-PRETRAIN-DATA-2026-0006

  • Claim: 当动作学习依赖多视图时,数据包应同步保存机位标识、视频、机器人状态和动作;10 Hz 是该 UR5 系统实例,不是预训练的通用帧率。
  • Stance: conditional | Confidence: direct
  • Paper: 2512.11612 Embodied Image Compression: Towards Codec for Robotic Visual Systems
  • Locator: Appendix C Subjective Data Collection
  • Evidence: 真实管线同步记录腕部与第三人称 RealSense、关节角和末端增量动作,频率为 10 Hz。
  • Quote: “Joint angles, two camera streams (wrist view and third-person view, captured by two Intel realsense cameras), and actions”
  • Authors: zhenghao-chen; zijie-yue; haozhe-li; et al.

EA-TWM-READ-0001

  • Claim: VT-WM 的训练序列同步记录腕部位姿、关节位置、外部视觉和两个指尖触觉视频,并使用时间戳对齐后降采样训练。
  • Stance: conditional | Confidence: direct
  • Paper: 2602.06001 Visuo-Tactile World Models
  • Locator: B.0.1 Training dataset
  • Evidence: 训练数据段明确列出了同步的本体状态、外部视频与双指触觉视频数据流。
  • Quote: “Each sequence contains multimodal data streams: proprioceptive information (wrist pose, joint positions), exocentric video from the camera, and video from each Digit 360 fingertip sensor. All data streams were synchronized using timestamps and downsampled to 6 FPS for training the world model. Our training dataset for V-WM and VT-WM consists of 124 demonstrations totaling 112k datapoints, with each demonstration averaging 40 seconds. For validation, we use 26 demonstrations spanning all tasks, c”
  • Authors: carolina-higuera; sergio-arnaud; byron-boots; et al.

EA-4D-READ-0012

  • Claim: 3PoinTr 保留含暂时遮挡点的轨迹,只屏蔽不可见的单个点—时间损失,因而能继续利用操作中任务关键的物体点监督。
  • Stance: conditional | Confidence: direct
  • Paper: 2603.08485 3PoinTr: 3D Point Tracks for Learning Manipulation from Unconstrained Human Videos
  • Locator: 4.3 Results: 3D Point Track Prediction
  • Evidence: 结果段对比了删除整条不可见轨迹的基线与仅屏蔽不可见 point-timestep 损失的 3PoinTr。
  • Quote: “The primary advantage of 3PoinTr is that it trains on data General Flow ignores. Real-world points are often temporarily occluded; General Flow removes any trajectory with invisible point-timestep pairs during preprocessing, whereas 3PoinTr retains all trajectories and masks losses for individual invisible point-timestep pairs. This provides additional supervision over task-critical object points that are temporarily occluded during manipulation. For example, in the Throw Away Paper task, every”
  • Authors: adam-hung; bardienus-pieter-duisterhof; jeffrey-ichnowski

EA-TWM-READ-0007

  • Claim: 触觉信息用于 VLA 或世界模型时,低频视觉语言理解和高频触觉反馈应在架构上分离。
  • Stance: conditional | Confidence: direct
  • Paper: 2605.07308 AT-VLA: Adaptive Tactile Injection for Enhanced Feedback Reaction in Vision-Language-Action Models
  • Locator: 5 Conclusion
  • Evidence: AT-VLA 把系统分为慢速视觉语言流和快速触觉流,慢速流负责任务理解和视觉定位,快速流以高频处理触觉反馈;作者采用 3:1 的快慢流频率比,并在真实接触丰富任务中验证 adaptive tactile injection、tactile gate、adaptive cross-attention 和 reaction dual-stream 的作用。
  • Quote: “In summary, AT-VLA introduces an adaptive framework that seamlessly integrates tactile sensing into vision-language-action models. Through the Adaptive Tactile Injection mechanism, AT-VLA dynamically balances pretrained visual-language knowledge with newly learned tactile representations, preserving model integrity while enhancing action precision. The Tactile Reaction Dual-Stream mechanism further enables rapid, high-frequency tactile responses by decoupling slow perceptual reasoning from fast”
  • Authors: xiaoqi-li; muhe-cai; jiadong-xu; et al.

EA-4D-READ-0010

  • Claim: HapTile 的数据质量控制在机器人控制环中同步全部模态,检查空轨迹、损坏轨迹和时间戳缺口,并验证动作—状态一致性。
  • Stance: conditional | Confidence: direct
  • Paper: 2606.04825 HapTile: A Haptic-Informed Vision-Tactile-Language-Action Dataset for Contact-Rich Imitation Learning
  • Locator: 3.2 Synchronization and Data Quality Control
  • Evidence: 数据质量段明确记录了控制环同步、时间戳缺口检查、损坏轨迹剔除和 action-state consistency 检查。
  • Quote: “All data modalities are synchronized through the robot control loop. For policy learning, actions are converted to a unified 7D end-effector delta representation (1) where are translational deltas, are rotational deltas, and is the gripper command. This decouples learning from the exact robot configuration, enabling cross-embodiment by focusing the policy on local contact adjustment from tactile feedback. Several quality checks are applied to every collected trajectory. Empty or corrupted trajec”
  • Authors: amirhosein-alian; yongqiang-zhao; shiyi-gu; et al.

EA-TWM-READ-0014

  • Claim: 在视觉遮挡或视觉退化下,显式把触觉接触投影到图像域可以改善空间对齐;但这种收益依赖对运动学和标定误差导致的空间不确定性建模。
  • Stance: conditional | Confidence: direct
  • Paper: 2606.08765 RGB-S: Image-Aligned Tactile Saliency for Robust Dexterous Manipulation
  • Locator: Abstract (full-text section)
  • Evidence: 作者称视觉观测不可靠或被遮挡时,稀疏异构触觉与稠密视觉表示的对齐是核心挑战;方法使用正运动学和相机标定投影触觉传感器位置,并用力调制高斯 saliency maps 建模运动学和标定误差带来的空间不确定性。
  • Quote: “We then render force-modulated Gaussian saliency maps to model spatial uncertainty arising from kinematic and calibration errors.”
  • Authors: shengcheng-luo

EA-PRETRAIN-DATA-2026-0001

  • Claim: 异构来源应扩大,但在联合预训练前必须将空间坐标、本体形态、物理时间和标签可靠性显式对齐或条件化;否则会降低动作学习性能。
  • Stance: conditional | Confidence: direct
  • Paper: 2606.17200 ACE-Ego-0: Unifying Egocentric Human and Robotic Data for VLA Pretraining
  • Locator: 5.2 Ablation Studies, Figure 5(b)
  • Evidence: 三项组件消融均降低 RoboCasa 成功率,其中去掉人类伪动作可靠性加权的降幅最大。
  • Quote: “Removing morphology tokens makes the success rate drop from 72.8% to 70.9%”
  • Authors: hao-li; ganlong-zhao; yufei-liu; et al.

EA-EGO-2026-0016

  • Claim: Ego-centric wrist trajectory 与相机自运动天然耦合;若不统一参考系,动作监督会把 camera rotation/translation 错算为 hand motion。
  • Stance: limit | Confidence: direct
  • Paper: 2606.06194 ActiveMimic: Egocentric Video Pretraining with Active Perception
  • Locator: 3 Method
  • Evidence: 方法段明确说明 current-frame wrist pose 与 first-frame camera path 的坐标差异会混合两类位移。
  • Quote: “using these wrist poses directly as action supervision would therefore conflate wrist movement with camera motion”
  • Authors: xingyao-lin; guojin-zhong; tianyi-lu; et al.

EA-VLABREAK-2026-0001

  • Claim: H-WM 用低频符号逻辑转移维持全局顺序,用潜在视觉子目标把逻辑状态落到感知空间,再由高频 VLA 执行动作 chunk。
  • Stance: support | Confidence: direct
  • Paper: 2602.11291 H-WM: Robotic Task and Motion Planning Guided by Hierarchical World Model
  • Locator: IV-C Hierarchical World Model Guidance for VLA
  • Evidence: 方法定义了逻辑世界模型、视觉世界模型、低层 VLA 和子任务完成检测的两时间尺度接口。
  • Quote: “The hierarchical information at multiple abstraction level enables the VLA to maintain consistency with long-horizon task structure while remaining responsive to local visual feedback.”
  • Authors: jinbang-huang; wenyuan-chen; zhiyuan-li; et al.

EA-ALIGN-READ-0004

  • Claim: Offline VLA indicators can fail to transfer to stable real-robot behavior when action semantics, coordinate frames, temporal modality alignment, image preprocessing, and dataset coverage are not controlled together.
  • Stance: limit | Confidence: direct
  • Paper: 2606.30456 Vision-Language-Action Models: Experimental Insights from a Real-World UR5 Platform
  • Locator: Abstract (full-text section)
  • Evidence: The UR5 study reports a gap between offline indicators and unstable closed-loop physical behavior, attributing it to data-model-control pipeline consistency rather than model capacity alone.
  • Quote: “Instead, it is strongly influenced by a combination of factors, including action semantics, coordinate frame conventions, temporal alignment between modalities, image preprocessing consistency, and dataset coverage and quality. These observations lead to a key interpretation: the successful deployment of VLA systems in real-world settings depends less on incremental improvements in model capacity and more on precise control of the entire data–model–control pipeline.”
  • Authors: mathilde-hochedel; marc-lalonde

EA-VLABREAK-2026-0007

  • Claim: 对 WAM 的安全监测不能只检查‘想象的未来是否看起来合理’,还必须验证未来与实际执行动作在闭环中是否同步。
  • Stance: limit | Confidence: direct
  • Paper: 2607.15207 BadWAM: When World-Action Models Dream Right but Act Wrong
  • Locator: 5.8 What Do These Results Imply for WAM Safety?
  • Evidence: 想象保持攻击在 40 个任务中有 39 个降低未来漂移,同时保留显著攻击强度。
  • Quote: “The relevant security property is not plausibility of the imagined future in isolation, but synchronization between the imagined future and the action that will actually be executed.”
  • Authors: qi-li; xingyi-yang; xinchao-wang

EA-TACTILE-2026-0001

  • Claim: 近一年触觉表征研究开始从小规模单任务管线走向大规模全手触觉—第一视角配对数据和多任务、任务级 OOD 基准;HT-Bench 以约 1000 万 RGB 帧、780 万触觉帧和 226 项任务测量接触结构、跨模态对齐与时间动态。
  • Stance: support | Confidence: direct
  • Paper: 2606.19161 HT-Bench: Benchmarking and Learning Dexterous Full-Hand Tactile Representations with Egocentric Vision
  • Locator: Abstract; 3 HT-Bench: A Multi-Task Tactile Evaluation Benchmark
  • Evidence: 摘要和基准设计章节直接给出数据规模、四项评测任务与任务级 OOD 划分。
  • Quote: “comprising 10M RGB frames and 7.8M tactile frames collected across 226 tasks.”
  • Authors: yuzhe-huang; jiaping-wu; jiaming-jiang; et al.

EA-SENSORERR-READ-0011

  • Claim: RGB-centric VLA 在照明变化导致的可见性退化下会暴露鲁棒性问题;事件流作为对照明更鲁棒、对运动敏感的补充观测,可以改善不同可见性水平下的动作预测。
  • Stance: support | Confidence: direct
  • Paper: 2606.29384 Event-VLA: Action-Conditioned Event Fusion for Robust Vision-Language-Action Model
  • Locator: Abstract (full-text section)
  • Evidence: 作者指出现有 VLA 往往假设稳定明亮的室内环境,而真实操作中 illumination shifts 会造成 degraded RGB observations;Event-VLA 将 degraded visibility 定义为 RGB-centric policies 的鲁棒性问题,并通过 action-query routing 将 event streams 融入 action representation,仿真和真实部署实验显示在不同可见性下保持更强鲁棒性。
  • Quote: “Abstract Vision-Language-Action (VLA) models have become an important paradigm of embodied AI. However, existing VLA models typically assume well-lit and stable indoor settings, while real-world embodied manipulation may involve degraded RGB observations caused by illumination shifts, posing critical challenges for robust robotic manipulation. To address this gap, we propose Event-VLA , an event-enhanced VLA framework for generalizable manipulation across varying illumination conditions. We form”
  • Authors: jiaxin-liu; xun-xu; zhenhao-zhang; et al.

References

Interactive Wiki

进入交互阅读

静态页采用默认版本;v0、v0、v0可在同一交互 Wiki 中阅读,不建立相互竞争的索引页。

Method and limits

研究方法、证据边界与引用说明

本页由当前结算的证据层自动派生,不改写已结算 run。论文只有在完整非 OCR 全文、结构化论文笔记和 claim-support audit 通过后,才进入接受证据。

页面结论是对所列论文与研究窗口的综合,不等同于无限外推。引用本专题时,请同时保留本页 canonical 地址、访问日期,并在需要时回溯原论文。