HiFi-UMI:用高保真UMI数据实现零机器人后训练
HiFi-UMI: Learning Deployable Manipulation Policies from High-Fidelity UMI Data Alone
HiFi-UMI造了一套便携数据采集系统,不用真机器人就能让策略直接部署,精度3mm,效果跟真实遥操作差不多,还开源了2000小时数据集,搞机器人操作的可以看看。
HiFi-UMI是一个便携式UMI数据生产系统,无需外部跟踪即可达到3毫米末端执行器精度。它通过头戴式立体惯性SLAM、原生相对位姿、微秒级GPIO触发和每只手两个广角相机(覆盖约200度)提升数据质量。
使用该数据后训练的策略可直接部署到真实机器人,在StarVLA-QwenPI、OpenPI-pi_0.5和LingBot-VA三个骨干上成功率差异仅为-2.
5、+3.1和-0.
6个百分点。在精密插入任务上最强策略达到85%成功率。
预训练4000小时可将十个未见任务的动作误差降低41%,并在StarVLA-QwenPI上将真实机器人成功提升18.1个百分点。
研究团队开源了2000小时微秒同步超广角数据集HiFi-UMI-2K。
HiFi-UMI: Learning Deployable Manipulation Policies from High-Fidelity UMI Data Alone
Learning deployable manipulation policies is bottlenecked by the scarcity of data that is both high-fidelity and scalable. Real-robot teleoperation is accurate but costly to scale; robot-free UMI capture scales readily, and current practice uses the resulting data mainly for pre-training, adding a small real-robot "anchor" at post-training. We ask whether raising the fidelity of robot-free UMI data, rather than shrinking the real-robot fraction, can remove that anchor. We present HiFi-UMI, a portable UMI data-production system co-designed for trajectory accuracy, inter-gripper relative pose, synchronization, and field of view: head-mounted offline stereo-inertial SLAM, native rather than reconstructed relative pose, a shared microsecond GPIO trigger, and two wide-angle cameras per hand covering ~200 degrees. It reaches 3 mm workspace-local end-effector accuracy without external tracking infrastructure. Using this corpus, we demonstrate zero-robot post-training: a policy post-trained solely on HiFi-UMI demonstrations deploys directly on a real robot and matches in-domain teleoperation across three backbones spanning the vision-language-action and world-action-model families, with success-rate differences of -2.5, +3.1, and -0.6 percentage points on StarVLA-QwenPI, OpenPI-pi_0.5, and LingBot-VA; the strongest policy reaches 85% on a precision insertion task, even though the teleoperation baseline is collected in the evaluation scene and no HiFi-UMI trajectory is. Pre-training on 4,000 hours from the same corpus lowers action error on ten unseen tasks by 41% and, on StarVLA-QwenPI, raises real-robot success by a further 18.1 percentage points. We open-source HiFi-UMI-2K, 2,000 hours of microsecond-synchronized, ultra-wide-FoV demonstrations, each automatically reconstructed and validated through simulation replay, as a large-scale, high-fidelity resource for the robot-learning community.
This report is categorized under AITOP's "Papers" desk. Key entities and technical topics under continuous tracking include HiFi-UMI, UMI, StarVLA-QwenPI, OpenPI.
Source attribution: curated from arXiv cs.LG (@Simple AI, :, Yuteng Wei, Jinming Ma, Jiawei Wang, Weitao Zhou, Yushen Zuo, Ke Rui, Minglei Li, Jinhao Zhang, Zhikang Pan, Xiang Wang, Haoran Jia, Huan Du, Zicheng Zeng, Jun Ma, Guiyu Qin, Di Zhang, Xiaofei Li).