The Year in AI Papers
The year/Independent research

Paper 2607.25895

HiFi-UMI: Learning Deployable Manipulation Policies from High-Fidelity UMI Data Alone

Simple AI, :, Yuteng Wei, Jinming Ma, Jiawei Wang, Weitao Zhou, Yushen Zuo, Ke Rui, Minglei Li, Jinhao Zhang, Zhikang Pan, Xiang Wang, et al.

Published
Jul 2026
Research lab
Independent
Citations
0
GitHub
Not linked

01 In brief

Summary

HiFi-UMI is a portable data-production system that co-designs hardware and software to achieve high-fidelity robot-free UMI data, aiming to eliminate the need for real-robot teleoperation in post-training.

It uses head-mounted offline stereo-inertial SLAM, native inter-gripper relative pose, microsecond GPIO synchronization, and ultra-wide 200° field-of-view cameras, achieving 3 mm end-effector accuracy.

The system has collected over 20,000 hours of data, with a curated 2,000-hour subset (HiFi-UMI-2K) released.

Across three backbones (StarVLA-QwenPI, OpenPI-π0.5, LingBot-VA), UMI-only post-training matches in-domain teleoperation, with success-rate differences of -2.5, +3.1, and -0.6 percentage points, respectively.

Pre-training on 4,000 hours reduces action error on unseen tasks by 41% and improves real-robot success by 18.1 percentage points.

The pipeline includes six stages: collection, reconstruction, simulation retargeting, AI-assisted annotation, human verification, and analysis/export, with a cumulative 96% usable-data yield.

The study demonstrates that high-fidelity robot-free data can serve as the sole post-training source for deployable policies, challenging the assumption that real-robot data is necessary for grounding.

02 From the paper

Abstract

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.