(Yicai) Dec. 2 -- China's rapidly growing industry of robots that look like humans requires standardization to mature, according to industry insiders.

"Money is not a problem for companies in the [humanoid robot] industry," Jiang Yuhua, general manager of Fosun Group’s master investment fund, said at a recent industry conference. In the past nine months, such firms have gained a cumulative investment of CNY4 billion (USD550 million). During the industry's initial development stage, opportunities abound for all, Jiang noted.

In the first 10 months of this year, the domestic industry completed at least 55 financing deals, according to the conference organizer GGII Robot Industry Research Institute. The institute predicts that China's annual sales of humanoid robots will exceed one million units by 2034.

Yicai learned of five additional deals last month, bringing the total from January to November to a minimum of 60 financing deals, with at least 22 exceeding CNY100 million (USD13.8 million) each.

Globally, around 150 commercial companies specialize in humanoid robots, with more than 80 based in China, said Xu Bin, general manager of Shanghai's National Local Joint Humanoid Robot Innovation Center. About half of these companies were founded by university students and teachers, Xu added.

Custom Hardware

However, developers lack unified standards in core hardware, creating challenges in improving the supply chain and research and development, Jiang from Fosun Capital explained. Moreover, manufacturers have not resolved algorithmic differences and disagree on the data that needs to be collected.

Shao Hui, general marketing manager at Juxie Intelligent Drive, told Yicai that customers' requirements are so customized and diverse that hardware suppliers struggle to meet the demand.

Some existing humanoid robot hardware, including bearings, remains "non-standard," said Niu Dongcheng, market director of Suzhou-based STT Extra Small Ball Bearings. This necessitates additional technical services from suppliers and complicates prompt technical support for distant customers.

Due to the lack of standards, robot manufacturers use disparate components, noted Fang Tingting, chairman of NIIC, a Suzhou-based real-time operating system firm. Sensors' access ways, communication, and interaction methods remain primitive and diverse, making it challenging to produce streamlined and efficient humanoid robots, Fang added.

The industry also faces fragmented application scenarios. Since data is held by individual customer companies, developers find it difficult to build universal training scenarios for humanoid robots, hindering successful commercial application, said Liu Yizhang, platform manager at Beijing's Embodied AI Robotics Innovation Center.

Editors: Tang Shihua, Emmi Laine