Skip to Content Facebook Feature Image

Simple AI Introduces HiFi-UMI, a High-Fidelity Robot-Free Data-Production System for Robot Manipulation Learning, with a 2,000-Hour Open Dataset

Business

Simple AI Introduces HiFi-UMI, a High-Fidelity Robot-Free Data-Production System for Robot Manipulation Learning, with a 2,000-Hour Open Dataset
Business

Business

Simple AI Introduces HiFi-UMI, a High-Fidelity Robot-Free Data-Production System for Robot Manipulation Learning, with a 2,000-Hour Open Dataset

2026-08-20 23:00 Last Updated At:23:15

A portable, high-fidelity robot-free data-production system for robot manipulation learning. Across three policy backbones evaluated in the report, policies post-trained solely on HiFi-UMI data reached success rates comparable to policies post-trained on in-domain real-robot teleoperation data.

NEW YORK, Aug. 20, 2026 /PRNewswire/ -- Simple AI has published the Tech Report for HiFi-UMI, a high-fidelity robot-free data-production system for robot manipulation learning, together with HiFi-UMI-2K, a 2,000-hour open dataset released under the Creative Commons Attribution 4.0 license. The Tech Report is available on arXiv and the dataset on Hugging Face.

Progress in robot manipulation learning is increasingly constrained by data. Real-robot teleoperation yields accurate, directly trainable trajectories but is difficult to scale: every hour of data requires the target robot, a teleoperation rig, and a skilled operator. Robot-free handheld demonstrations are cheaper and easier to scale, but have primarily been used for pre-training. Task-specific post-training, the stage that grounds a policy for real-robot deployment, has typically continued to rely on a smaller amount of real-robot teleoperation data as an anchor.

The report examines whether raising the fidelity of robot-free demonstration data, rather than shrinking the real-robot fraction, can remove that anchor for target-task post-training.

HiFi-UMI is a portable data-production system co-designed for four fidelity properties. Pose accuracy comes from head-mounted offline stereo-inertial SLAM, which the report measures at 3mm workspace-local end-effector accuracy. Cross-sensor timing is aligned to below 40 microseconds through a shared hardware trigger across all cameras and sensors. Inter-gripper relative pose is measured natively rather than reconstructed. Per-hand sensing covers approximately 200 degrees of field of view through two non-parallel wide-angle fisheye cameras. Every captured demonstration passes through automatic trajectory reconstruction and simulation replay validation, each gate with an approximately 98% pass rate.

The report evaluates the approach across three policy backbones spanning the vision-language-action and world-action-model families, and four bimanual tabletop tasks. Policies post-trained solely on HiFi-UMI demonstrations reached success rates comparable to policies post-trained on in-domain real-robot teleoperation data, with reported differences of −2.5, +3.1, and −0.6 percentage points across the three backbones. On a precision insertion task, the strongest HiFi-UMI-only policy reached 85% success under conditions where the teleoperation baseline had the advantage of being collected in the evaluation scene. Separately, pre-training on 4,000 hours of the same corpus reduced offline action prediction error on ten unseen tasks by 41%, and increased real-robot success on one of the evaluated backbones by 18.1 percentage points at matched post-training data.

"We wanted to test whether fidelity, rather than scale alone, is what unlocks robot-free data for deployment-oriented training," said Xiaofei Li, founder of Simple AI. "The report shows what this can look like within a specific set of tasks and models. By open-sourcing HiFi-UMI-2K, we hope to give the wider research community a shared, high-fidelity resource for continuing to study this question."

The report characterizes these findings as approximate aggregate parity within the tested models, tasks, and experimental conditions. The deployment robot uses the same gripper and wrist-camera configuration as the capture setup, with the main embodiment difference being robot arm kinematics. The comparison is not sample-matched, with 3,200 HiFi-UMI trajectories per task set against approximately 300 teleoperation trajectories, and reflects a comparison between practical data-production pipelines rather than a claim of per-trajectory equivalence. The report does not generalize the result to all robot learning settings, and does not conclude that real-robot data is no longer required in the broader field.

HiFi-UMI-2K is distributed in a training-ready format with synchronized multi-view video, bimanual end-effector trajectories, gripper states, language annotations, and subtask boundaries. Human faces in the recordings are masked before release. The paper reached No. 1 on Hugging Face Daily Papers on July 29.

HiFi-UMI is one component of Simple AI's work across foundation models, high-fidelity data, robotic systems, and real-world deployment. The company welcomes conversations with research groups and industry partners interested in high-fidelity data for robot learning.

About Simple AI
Simple AI is an embodied AI company developing general-purpose embodied intelligence systems for human living spaces. Its work integrates foundation models, high-fidelity data, robotic systems, and real-world deployment across the full embodied AI stack.

Resources

Tech Report: arxiv.org/abs/2607.25895 

Dataset: huggingface.co/datasets/simple-world-lab/HiFi-UMI-2K 

Project page: cloud.simpleai.tech/simple-world-lab/hifi-umi/ 

Media Contact

Grant Xin

Simple AI

media@simpleai.tech

** This press release is distributed by PR Newswire through automated distribution system, for which the client assumes full responsibility. **

Simple AI Introduces HiFi-UMI, a High-Fidelity Robot-Free Data-Production System for Robot Manipulation Learning, with a 2,000-Hour Open Dataset

Simple AI Introduces HiFi-UMI, a High-Fidelity Robot-Free Data-Production System for Robot Manipulation Learning, with a 2,000-Hour Open Dataset

YIWU, China, Aug. 20, 2026 /PRNewswire/ -- How much meaning can a single greeting card hold? Far more than paper and ink, it can carry affection across borders and connect people through shared emotions. For centuries, greeting cards have been bearers of human connection. That particular romance does not pale across national and cultural boundaries. Recently, an overseas Chinese blogger drew widespread attention with a video showing greeting cards prominently displayed in shopping malls abroad—and selling at surprisingly high prices. Long-established yet relatively niche, the greeting card industry is now being reshaped by changing patterns of global commerce.

"True" Dried Flowers – Preserving Two Decades of Memories

Yiwugo.com, the official website of the Yiwu Commodity Market, is the largest commodity wholesale market in the world. Founded in 1995, Yihenuo Stationery Firm has been with Yiwugo for 15 years. The company has evolved from a domestic-focused business into a cross-border exporter, witnessing the full arc of change in the greeting card industry over the decades. Among its offerings are timeless classics that have remained popular for 20 years, moving hundreds of thousands of cards each month, as well as bestselling pop-up cards that bring in several million yuan in annual sales.

As the person in charge of Yihenuo's operations in Yiwu, Zhu Liying carries much of the weight of its domestic and international business. Over more than three decades, Yihenuo has navigated rising raw material costs, industry-wide upgrades in equipment and processes, advances in printing technology, and evolving consumer preferences for greeting cards. Through it all, the company has retained the vast majority of its customers by delivering reliable quality, thoughtful services, and products that bring buyers back time and again.

According to Zhu Liying, one of the factory's signature products is a greeting card featuring real dried flowers on the cover. Introduced nearly 20 years ago, it has seen little change in price over the years. A mature dried-flower processing technique gives the cover a rich, three-dimensional texture and distinctive visual appeal. Since its launch, the card has drawn virtually no complaints about mold or flowers coming loose, helping it remain highly regarded by both new and longtime customers. The card is particularly popular in Japan and South Korea. Because the design is not tied to any specific holiday or occasion, it appeals to a broad range of consumers. Combined domestic and overseas sales have consistently reached hundreds of thousands of cards per month, making the product a key contributor to Yihenuo's strong reputation in the market.

From Card-by-Card Inspection to Inspection-Free: A Decade of Trust

As consumer tastes evolve, a growing number of buyers are drawn to intricately crafted pop-up cards that serve both as heartfelt greetings and as decorative pieces. Riding this trend, Yihenuo has rolled out a range of new designs, including pop-up cards and gift sets pairing cards with notebooks and blind boxes. Through Yiwugo, these products continue to attract high-quality buyers from around the world.

A well-known international brand with thousands of chain stores first connected with Yihenuo through Yiwugo 10 years ago. Initially, the client placed only small trial orders and inspected every item individually. After one to two years of rigorous evaluation, Yihenuo was granted inspection-free supplier status with the brand. Today, Yihenuo supplies the customer with nearly RMB 10 million worth of pop-up greeting cards annually, including customized products featuring major licensed IPs like Disney.

Handcrafted "Butterfly" Cards Take Flight - With a Million Sold Per Month

Ying Yuping's Kalatu Stationery Firm, on the other hand, specializes in premium handcrafted greeting cards. A brand with 28 years of history, Kalatu began connecting with overseas buyers through Yiwugo eight years ago. Its products are now sold across Europe, the US and the Middle East, as well as top-tier Chinese cities.

According to Ying, pop-up cards require not only sophisticated structural design but also skilled hand assembly. Paper quality and the precision of production equipment should also meet exacting standards, keeping unit costs relatively high. Yet the intricate designs and handcrafted feel have proved a hit with overseas buyers, pushing monthly sales to 1 million cards during peak seasons.

In 2024, Kalatu launched a versatile butterfly-themed greeting card suitable for occasions such as Mother's Day, Valentine's Day and Teachers' Day, both in China and abroad. The product quickly became a bestseller, generating more than RMB 1 million in sales in a single month and proving particularly popular in Europe and the US, where it secured strong repeat orders.

A greeting card may be small, but within its few square inches lies a vast global market built on human connection. Such cards link not only one person's feelings with another, but also a traditional industry with new opportunities in a changing era. From timeless classics to imaginative pop-up designs, greeting-card makers rooted in Yiwu continue to refine their products while leveraging Yiwugo to bridge domestic and international trade channels, bringing Chinese papercraft to consumers around the world. The industry may see its share of ups and downs, but by staying focused on quality and responsive to evolving market needs, even a niche business can find ample room to grow in global trade.

** This press release is distributed by PR Newswire through automated distribution system, for which the client assumes full responsibility. **

Small Cards, Big Emotions, Global Reach

Small Cards, Big Emotions, Global Reach

Recommended Articles