TMTPOST--Marc Raibert, the founder of Boston Dynamics and often referred to as the "father of the robotic dog," made his first visit to Beijing for the 2024 World Robot Conference. He expressed his amazement at the event, saying that he can hardly believe not only the number of robots but also the incredible enthusiasm of the people attending the exhibition. China's excitement and passion for robots are incredibly high.

The four CEOs who conversed with Marc are from well-known companies as well. Apart from Unitree Robotics, the other three companies, namely Robotera, GalaxyBot, and the Beijing Embodied Intelligence Robotics Innovation Center, were all incorporated less than two years ago, yet they have quickly become leading names in the fields of embodied intelligence and humanoid robots.

Behind Robotera and GalaxyBot, there also stand the most prestigious institutions of higher learning in China: Peking University and Tsinghua University, representing an important source of China's cutting-edge technological innovation.

Robotera was incubated and established by the Institute of Cross-disciplinary Information Research of Tsinghua University led by Yao Qi, and the founder of Robotera, Chen Jianyu, is still an assistant professor at the Institute of Cross-disciplinary Information Research of Tsinghua University. The company has raised more than ten million yuan in financing, with Lenovo Venture Capital being one of the investors.

The founder of GalaxyBot Robot, Wang He, is an assistant professor at Peking University and also the director of the Embodied Intelligence Research Center of Beijing Zhiyuan Artificial Intelligence Research Institute. In the past six months, GalaxyBot has raised 700 million yuan in angel financing round, setting a new record for the single round of financing scale in the domestic humanoid robot field. Meituan Investment, BAIC Investment, Sensetime Guoxiang Fund, iFLYTEK Fund, Qiming Venture Capital, Blue Run Venture Capital, Matrix Venture Capital, Yuanma Capital, IDG Capital and other top financial institutions have all jumped on the bandwagon.

And the Beijing Embodied Intelligence Robot Innovation Center is a "national team" institution for robots initiated by the Beijing Municipal Government. Xiong Youjun, the general manager of the Beijing Embodied Intelligence Robot Innovation Center, is currently the CTO of Ubtech Technology, the "first stock of humanoid robots."

Unitree Robotics is also a star company in the humanoid robot industry. Wang Xingxing, the company's founder, CEO, and CTO, graduated from Shanghai University with a master's degree in mechanical engineering, and has nearly more than ten years of experience in the development of legged robots. He possesses the capability to develop and commercialize products from scratch to completion, holding over ten patents related to robotics.

The roundtable was moderated by Professor Alois C.Knoll from the Technical University of Munich. The discussion featured Marc Raibert, the founder of Boston Dynamics, alongside leading figures from China's top AI robotics companies: Chen Jianyu, Wang Xingxing, Wang He, and Xiong Youjun. The five discussed topics including the robotics industry and the impact of AI large models on robotics development.

During the hour-long discussion, Raibert emphasized that while hardware development in robotics will remain crucial over the next five years, foundational models and other AI methods that break down semantic understanding barriers will also have a significant impact on robotics technology. However, achieving real progress in the field will require greater reliability and safety, necessitating substantial resource investment to address these challenges.

Raibert expressed skepticism about the humanoid robotics industry. He believes that while industrial robots are highly advanced and commercially viable, humanoid robots are, to some extent, more of a showcase than a practical tool for productivity.

Chen said that in the past ten years, AI has developed very quickly, from simple image processing to deep learning, reinforcement learning, and then to ChatGPT technology, which has a great impact on the future "embodied intelligence". In the next few years, AI and embodied intelligence will continue to integrate with humanoid robots to train a general model. He believes that in the next five years, human beings will usher in the "ChatGPT moment of robots".

Wang noted that AI large models still face significant limitations when applied to robotics, particularly with skills like "generalized bottle-opening," which have not yet been fully developed. However, he also sees this as a major opportunity, suggesting that as the capabilities of large models continue to improve, there will be potential for general-purpose robots to handle operational tasks within the next five years.

Xiong said that advancements in joint performance, high energy, and battery density are rapidly improving robotics. These developments significantly enhance robots' motion control and motor capabilities. Over the next five years, improvements in robotic joints and energy density are expected to continue. Despite this progress, there remains a significant gap in energy efficiency between robots and humans.