It is estimated that humanoid Bots will take at least 10 years to land?

The Science and Technology Innovation Board Daily reported on May 21 (Reporter Li Mingming) that as an important carrier of AI landing in the physical world, discussions about the bubble of humanoid robots continue amid the ongoing influx of capital.

At the 2025 Sohu Technology Annual Forum hosted by Sohu, topics of public concern such as the current status and controversies of commercialization, talent cultivation, and more were discussed. Zhao Mingguo, a researcher from Tsinghua University's Department of Automation and director of the Robot Control Laboratory, Chang Lin, founder and CEO of Leju Robotics, Zhang Zhengtao, researcher at the Institute of Automation, Chinese Academy of Sciences and founder and chairman of Lingbao CASBOT, and Wang Qizhou, deputy general manager of Yushu Technology shared their different perspectives.

Chang Lin stated that whether humanoid robots have a bubble depends on whether they are useful, whether they can solve problems in production and life, and create value. The Yizhuang Marathon bringing robots into reality is a necessary path for them to enter the public, and this core issue should not be avoided.

Zhao Mingguo believes that the development of humanoid Bots requires long-term accumulation, similar to autonomous driving which takes a 20-year cycle, and conservatively estimates it will take at least 10 years. "The public should not have such high expectations that tomorrow Bots can walk into homes like buying a television or refrigerator; these activities need to be aligned."

Zhang Zhengtao introduced his views from the perspectives of interdisciplinary integration and industrial application. He believes that humanoid Bots are a large device generated from multiple disciplines, which is what drives its value.

On the implementation of the industry, Zhang Zhengtao said that the industry is open to humanoid robots, and the current independent decision-making ability of the whole machine is insufficient, so it needs to be combined with traditional visual servo control through remote operation, hand-eye collaboration and traditional visual servo control. With the accumulation of data, there will be qualitative changes in this field in the next 1-3 years, and the form of qualitative change is not necessarily bipedal, and humanoid or wheel-arm robots may also become the breakthrough direction.

Wang Qizhou stated that Yushu has proven itself over nine years, transitioning from having no clients and focusing solely on research and development for universities to now being applied in fields such as electricity and chemicals.

When discussing changes in capital sentiment and the financing environment, both Chang Lin and Zhang Zhengtao felt a significant turning point. Chang Lin pointed out that since 2022, when the country positioned humanoid Bots as the "next generation of disruptive products after PCs, smartphones, and new energy vehicles," the policy support has continued to strengthen, and local governments view it as a high point for driving the industrial chain, with various regions scrambling to layout.

When it comes to talent cultivation for humanoid Bots, Zhao Mingguo and Wang Qizhou shared their respective views from academic and industrial perspectives.

Wang Qizhou revealed, "Wang Xingxing mentioned a few days ago at a meeting that Yushu is very short of personnel in all aspects. Initially, we discussed that humanoid Bots involve interdisciplinary fields, including ontology, AI, perception, and even some cutting-edge new materials, all of which require investment in human resources for research and development to achieve a relatively good state."

He stated that Yushu has preferred talents over the years who have some hands-on abilities, are more willing to try new technologies, learn, conduct research and development, and even participate in competitions.

Zhao Mingguo believes that the explosion in the field of humanoid robots is set against the backdrop of AI, with embodied intelligence becoming an important direction. China has a significant talent advantage in this area, as 70% of the global research talent in the robotics field is of Chinese descent, and those who are most competitive in embodied intelligence are also primarily Chinese.

In addition, he also mentioned that domestic low-cost Bots platforms (such as Yushu products) have formed an "unlimited firing power" advantage in trial and error, lowering the experimental threshold and accelerating talent cultivation.

However, he also pointed out that there is a problem in the country of not being very good at doing 0 to 1, leaning towards follow-up innovation from 1 to 10, and lacking original breakthroughs in the deep thinking aspects of underlying technologies such as pulse neural networks and general AGI, thus requiring a strengthening of basic theoretical research.

Source: Science and Technology Innovation Board Daily

Author: Kexin Board Daily