WUWENAI Launches Physical AI Data Infrastructure Platform “Wuyin” in Zhejiang

Physical AI data infrastructure company WUWENAI has launched the industry’s first physical AI data foundation platform, named Wuyin, in Deqing, Zhejiang province.

The launch event attracted more than 50 robotics and AI companies, including Horizon Robotics, Xingdong Jiyuan, Lingxin Dexterous Hand, Deep Robotics and Digu Robotics.

The Wuyin platform introduces the concept of a “data foundation layer”, built around three core capabilities:

• a high-quality data system
• a high-value scenario ecosystem
• a Real2Sim2Real closed-loop toolchain

The data system integrates real-world data collection with high-fidelity simulated datasets. The scenario ecosystem recreates real application environments using simulation technologies, while the toolchain connects data acquisition, model training, scenario evaluation and industrial deployment.

The platform has already accumulated more than 1,000 TB of data and has open-sourced 10,000 hours of high-quality datasets.

For data collection, Wuyin supports multiple methods including VR teleoperation, exoskeleton operation, UMI capture, first-person human data recording and motion capture. Cross-embodiment data transfer tools also enable different robots to reuse the same dataset.

In addition, the platform uses generative simulation technologies to build a library of millions of “sim-ready” assets, allowing real-world environments to be mapped into virtual simulations through a Real2Sim2Real workflow. Current application scenarios include logistics warehousing, industrial manufacturing, home services, retail, hospitality and tourism, and office automation.

WUWENAI was founded in 2022 by Liu Shengxiang, former head of autonomous-driving data infrastructure at Baidu Apollo. The company aims to address the embodied AI industry’s key bottlenecks—high data-collection costs and limited scalability—by building third-party data infrastructure.

Source: QbitAI