8 11, openThe Wuhan Working Group of the OpenLET Embodied AI Open Source Dataset Community (hereinafter referred to as "Wuhan Working Group") was officially launched at Optics Valley., marksHubei's first national-level open-source data collaboration platform for embodied AI officially launches in Wuhan.
The Wuhan Working Group of the OpenAtom OpenLET Embodied AI Open Source Dataset Community was officially launched in Optics Valley.
According to reports,OpenLET is a national-level open-source dataset community for embodied AI.Initiated by the OpenAtom Foundation and led by Leju Robotics, this initiative focuses on open-sourcing real-world humanoid robot data, sharing toolchains, and cultivating a developer ecosystem. It has already released massive amounts of real-machine data, with industry-leading download volumes across the web.
"The humanoid robotics industry has reached a critical bottleneck: while hardware and basic motion control technologies are mature, the 'robot brain' responsible for decision-making lacks sufficient real-world operational data for training and optimization." Dr. Wang Song, Technical Director of the OpenLET Embodied AI Open Dataset Community who traveled to Wuhan, explained that datasets are the core foundational resource for developing the embodied AI industry in Wuhan. Unlike general-purpose large models, humanoid robots rely on physical interactions to walk, grasp, and perform industrial tasks; high-quality real-world interaction data serves as the essential curriculum for their iterative evolution.
Dr. Wang Song, Chair of the Technical Committee for the OpenLET Embodied AI Open Dataset Community
Wang Song explained that in the past, each company kept its robot operation data, motion parameters, and scenario experience locked on their own hard drives, creating "data silos." Open sourcing means sharing these valuable real-world robot operation datasets and training tools so university research teams and small-to-medium enterprise developers can freely conduct research and train algorithms without duplicating costly data collection efforts. The "big move" under OpenLET involves releasing over 6 minutes of full-size humanoid robot real-world operation data, covering action samples, environmental perception, and interaction behaviors across real-life scenarios such as homes, factories, and commercial spaces. This enables research teams and small-to-medium robotics companies to use the data out-of-the-box while jointly establishing unified industry standards for data collection, labeling, and circulation.
Wang Song introduced,Leveraging open-source datasets and integrated development tools, Wuhan-based automotive, optoelectronics, and equipment manufacturing enterprises can rapidly train humanoid robots tailored to their production lines, accelerating intelligent transformation across the manufacturing sector.
Just last month (7), Wuhan proposed at the conference on advancing first-tier city construction in digital intelligence economy to accelerate the flow of public data. By 2028, the city aims to build more than 200 high-quality industry datasets, supporting the training of over 200 vertical large models and agent development.
The newly established Wuhan Working Group is led by Bai Xiang, Dean of the School of Software at Huazhong University of Science and Technology, as Chair. The Wuhan Optics Valley Construction Investment Co., Ltd. serves as Executive Vice-Chair, with Dai Yangchun, Member of the Party Committee and Deputy General Manager, representing the company at the launch ceremony.The working group brings together nearly 70 universities and research institutions in the province, including Huazhong University of Science and Technology, Wuhan University, Wuhan University of Technology, and Central China Normal University, as well as over 30 enterprises across the industrial chain.Wuhan Optics Valley Construction Investment Co., Ltd. plans to leverage the industry-education integration training base to establish the "Central China Embodied Robot Intelligent Connection and Scenario Coordination Center," supporting the Wuhan working group in carrying out relevant tasks.
Working groups have been established in Hefei, Beijing, Foshan, Shanghai... while WuHanyang possesses strong university research resources, a complete smart manufacturing industry, and an active cluster of tech startups, making it the core hub for OpenLET's central China ecosystem."Wang Song explained that after the Wuhan team is established, it will share standardized real-device and simulated multimodal open-source datasets across the entire region. Local enterprises can directly leverage these mature data systems to catch up with industry frontiers. The team will also collaborate with over 100 universities and colleges province-wide, opening up standardized datasets to support faculty-student research projects and academic publications while meeting the urgent needs of local intelligent manufacturing transformation. In the future, leveraging the Guanggu Industrial Park and manufacturing scenarios, the initiative will drive validation and deployment of embodied AI technologies in industrial manufacturing, warehousing logistics, and park services, accelerating intelligent upgrades for the real economy.