Strategies for Speeding Up the Commercialization of Humanoid Robots
3 April 2025 · Uncategorized ·
Source: · https://www.cnr.cn/tech/techxp/20240914/t20240914_526903055.shtml
A mechanical hand flexibly mimics gestures made by audience members, performing actions such as grabbing and extending fingers. Similarly, a humanoid robot that is about the same height as an average adult can climb slopes, stairs, and obstacles. On automobile production lines, these robots are attempting to complete complex tasks like flexible assembly and product inspection.
At the recently held 2024 World Robot Conference, more than 600 innovative products from 169 companies were showcased, with several humanoid robots stealing the show. Ten years ago, humanoid robots could only perform simple limb control; today, some of them have intelligent 'brains,' agile 'little brains,' and flexible limbs.
With continuous advancements in technology and a faster formation of upstream and downstream industrial chains, commercial applications for humanoid robots are expected to accelerate significantly. Robots Are Becoming More Human-Like
Last year at the end of 2023, UBTECH went public, becoming China's first publicly traded 'humanoid robot company'; this year, Tesla unveiled its second-generation Optimus (Optimus). The热度持续攀升of humanoid robots has continued to rise.
Advancements in technology are making machines more human-like. According to Qiao Hong, president of the World Robot Cooperation Organization and an academician at the Chinese Academy of Sciences, dedicated components and materials for humanoid robots have become increasingly popular; high-explosive generators, precision reducers, highly accurate sensors, long-lasting batteries, etc., will form a stable and high-performance hardware system. Additionally, based on neural networks, graph grammars, evolutionary algorithms, among other artificial intelligence technologies, humanoid robots can automatically construct leg-foot modules, arms, trunks according to scene and task requirements, achieving coordinated optimization of morphology and control.
'Previously, robots lacked autonomous motion control capabilities; they could only complete single tasks in fixed environments,' said Xiong Youjun, general manager of the Beijing Embodied Intelligence Robot Innovation Center. 'Nowadays, large models have enabled humanoid robot brains to achieve human-machine interaction through multimodal means.'
Cao Wei, a partner at BlueRun Ventures, says that recent applications of big models have had significant impacts on the robotics industry. Traditional algorithms generally perform poorly in complex task execution; however, introducing these new technologies has significantly improved robots' operational capabilities and greatly increased their potential for commercialization.
Xing Xingxing, founder of Unitree Robotics, also believes that over the past decade, deep reinforcement learning and large language models have profoundly influenced humanoid robot development. In the next few years, both are expected to make greater progress and further integrate with humanoid robots.
Human-Robot Complementarity Is a Trend
Will scenarios where machines replace humans move from science fiction into reality?
In Xiong Youjun's view, humanoid robots will not replace human beings; instead, they may form complementary relationships that enhance production efficiency and safety. In industrial settings, future factories might adopt the '721' model: 70% of tasks would be handled by currently widely used specialized robots such as robotic arms or unmanned transport vehicles; another 20%, complex tasks beyond these machines’ capabilities, will likely fall to humanoid robots that can enter environments better suited for human interaction. The remaining 10% might still require manual intervention due to technical limitations.
Xie Ning, chief professor of robotics and automation at the University of Hong Kong and director of its Emerging Technology Institute, believes that introducing humanoid robots is mainly about completing tasks humans are unwilling or unsuitable to do, thereby further promoting industry automation and intelligence.
'Although they resemble people in appearance, from a practical standpoint, industries hope these machines can become super-limit robots capable of doing more work beyond human capabilities,' Xie said. 'For example, there's still significant room for development regarding flexibility, stability, load-bearing capacity, working hours.'
Commercialization Requires Overcoming Challenges
The rapid evolution of humanoid robot technology has made it a new high ground in technological competition and an emerging industry track with broad economic potential. In October 2023, the Ministry of Industry and Information Technology issued 'Guiding Opinions on Promoting Innovative Development of Humanoid Robots,' setting goals for China's humanoid robotics sector: by 2025 to establish a preliminary innovation system; by 2027 significantly enhance technological capabilities so that overall strength reaches world-leading levels.
China’s rich industrial ecosystem and consumer scenarios, vast market space, along with advancements in technologies like big models provide strong support for the rapid development of humanoid robots. According to recent statistics from China's Ministry of Industry and Information Technology, over the past three years new installations of industrial robots have accounted for more than half globally; manufacturing robot density has reached 470 units per ten thousand workers, nearly a nineteen-fold increase in just one decade.
However, it must be acknowledged that currently commercialization is still at an early stage. What challenges do humanoid robots need to overcome before they can enter households?
One challenge lies with functionality: some insiders admit the current upper limb precision of humanoid robots is around centimeters which lags far behind industrial-grade accuracy measured in 0.01 millimeters. The 'Guiding Opinions' propose that by 2025 a series of key technologies for humanoid robot brains and limbs should be broken through to ensure safe and effective supply of core components.
Another challenge is cost: currently China's market has seen some lightweight low-cost products like Unitree Robotics’ G1 mass-produced version priced at RMB9.9 million (approximately USD 1.45 million) or more; however, most humanoid robots on the market are priced in tens to hundreds of thousands of dollars.
'From a supply chain perspective,' said Xiong Rong chief scientist at Zhejiang Humanoid Robot Innovation Center 'to break through core component manufacturing challenges and achieve cost reduction we still rely heavily on China's industrial foundation.'
At the recently held 2024 World Robot Conference, more than 600 innovative products from 169 companies were showcased, with several humanoid robots stealing the show. Ten years ago, humanoid robots could only perform simple limb control; today, some of them have intelligent 'brains,' agile 'little brains,' and flexible limbs.
With continuous advancements in technology and a faster formation of upstream and downstream industrial chains, commercial applications for humanoid robots are expected to accelerate significantly. Robots Are Becoming More Human-Like
Last year at the end of 2023, UBTECH went public, becoming China's first publicly traded 'humanoid robot company'; this year, Tesla unveiled its second-generation Optimus (Optimus). The热度持续攀升of humanoid robots has continued to rise.
Advancements in technology are making machines more human-like. According to Qiao Hong, president of the World Robot Cooperation Organization and an academician at the Chinese Academy of Sciences, dedicated components and materials for humanoid robots have become increasingly popular; high-explosive generators, precision reducers, highly accurate sensors, long-lasting batteries, etc., will form a stable and high-performance hardware system. Additionally, based on neural networks, graph grammars, evolutionary algorithms, among other artificial intelligence technologies, humanoid robots can automatically construct leg-foot modules, arms, trunks according to scene and task requirements, achieving coordinated optimization of morphology and control.
'Previously, robots lacked autonomous motion control capabilities; they could only complete single tasks in fixed environments,' said Xiong Youjun, general manager of the Beijing Embodied Intelligence Robot Innovation Center. 'Nowadays, large models have enabled humanoid robot brains to achieve human-machine interaction through multimodal means.'
Cao Wei, a partner at BlueRun Ventures, says that recent applications of big models have had significant impacts on the robotics industry. Traditional algorithms generally perform poorly in complex task execution; however, introducing these new technologies has significantly improved robots' operational capabilities and greatly increased their potential for commercialization.
Xing Xingxing, founder of Unitree Robotics, also believes that over the past decade, deep reinforcement learning and large language models have profoundly influenced humanoid robot development. In the next few years, both are expected to make greater progress and further integrate with humanoid robots.
Human-Robot Complementarity Is a Trend
Will scenarios where machines replace humans move from science fiction into reality?
In Xiong Youjun's view, humanoid robots will not replace human beings; instead, they may form complementary relationships that enhance production efficiency and safety. In industrial settings, future factories might adopt the '721' model: 70% of tasks would be handled by currently widely used specialized robots such as robotic arms or unmanned transport vehicles; another 20%, complex tasks beyond these machines’ capabilities, will likely fall to humanoid robots that can enter environments better suited for human interaction. The remaining 10% might still require manual intervention due to technical limitations.
Xie Ning, chief professor of robotics and automation at the University of Hong Kong and director of its Emerging Technology Institute, believes that introducing humanoid robots is mainly about completing tasks humans are unwilling or unsuitable to do, thereby further promoting industry automation and intelligence.
'Although they resemble people in appearance, from a practical standpoint, industries hope these machines can become super-limit robots capable of doing more work beyond human capabilities,' Xie said. 'For example, there's still significant room for development regarding flexibility, stability, load-bearing capacity, working hours.'
Commercialization Requires Overcoming Challenges
The rapid evolution of humanoid robot technology has made it a new high ground in technological competition and an emerging industry track with broad economic potential. In October 2023, the Ministry of Industry and Information Technology issued 'Guiding Opinions on Promoting Innovative Development of Humanoid Robots,' setting goals for China's humanoid robotics sector: by 2025 to establish a preliminary innovation system; by 2027 significantly enhance technological capabilities so that overall strength reaches world-leading levels.
China’s rich industrial ecosystem and consumer scenarios, vast market space, along with advancements in technologies like big models provide strong support for the rapid development of humanoid robots. According to recent statistics from China's Ministry of Industry and Information Technology, over the past three years new installations of industrial robots have accounted for more than half globally; manufacturing robot density has reached 470 units per ten thousand workers, nearly a nineteen-fold increase in just one decade.
However, it must be acknowledged that currently commercialization is still at an early stage. What challenges do humanoid robots need to overcome before they can enter households?
One challenge lies with functionality: some insiders admit the current upper limb precision of humanoid robots is around centimeters which lags far behind industrial-grade accuracy measured in 0.01 millimeters. The 'Guiding Opinions' propose that by 2025 a series of key technologies for humanoid robot brains and limbs should be broken through to ensure safe and effective supply of core components.
Another challenge is cost: currently China's market has seen some lightweight low-cost products like Unitree Robotics’ G1 mass-produced version priced at RMB9.9 million (approximately USD 1.45 million) or more; however, most humanoid robots on the market are priced in tens to hundreds of thousands of dollars.
'From a supply chain perspective,' said Xiong Rong chief scientist at Zhejiang Humanoid Robot Innovation Center 'to break through core component manufacturing challenges and achieve cost reduction we still rely heavily on China's industrial foundation.'