In 1 to 2 years, robots will autonomously complete simple dangerous tasks. In 3 to 5 years, robots will perform complex hazardous work on their own, with humans supervising only in exceptional situations.
Yan Weixin, chief scientist at the Shanghai Jiao Tong University Institute of Artificial Intelligence (AI), told ChosunBiz on Aug. 4 that this is when Humanoid Robot could replace humans in dangerous manufacturing sites.
Yan designed Shanghai's AI development plan and served as a Commissioner at a key laboratory of the Ministry of Industry and Information Technology (MIIT), which oversees China's advanced industry policy. The Shanghai Jiao Tong University AI Institute, where Yan currently works, is a key forward base for China's AI policy that plans Shanghai's AI development blueprint and carries out state-led special-purpose robot projects.
AgiBot, the robotics company co-founded by Yan, shipped 5,168 Humanoid Robot units last year, ranking No. 1 globally.
Just over 10 years ago, China lingered on the fringes of industrial robots. Now it is leading trends in the global Humanoid Robot market. Moving beyond the past model that relied on massive subsidies, China densely clustered the value chain of key components in vast industrial parks, achieving overwhelming cost competitiveness.
On top of that, a distinctive culture of innovation—training robots freely with wild data from real industrial sites where variables abound, rather than in controlled greenhouse laboratories—is accelerating rapid evolution.
There are currently 160 Humanoid manufacturers in China, and nearly 10,000 corporations engaged in robot-related businesses, including 600 key component suppliers. The following is a Q&A with Yan.
—What explains the explosive growth of China's robot ecosystem?
Powerful collaboration synergies across the entire value chain, dense supply networks, and a willingness to invest in basic research. In massive advanced manufacturing clusters such as the Yangtze River Delta centered on Shanghai and the Pearl River Delta around Shenzhen, every part of the supply chain—molding, injection, precision machining—can be sourced locally.
This concentrated infrastructure has driven groundbreaking cost reductions. Another key is training algorithms with vast, living data obtained from chaotic real-world manufacturing and everyday environments.
—What are the weaknesses?
We have achieved about 90% domestic production and stable supply for structures such as robot bodies and about 80% for batteries, but reliance on imports for high-end core components remains high.
The domestic production rate for torque sensors, which realize delicate sensing, is under 10%, and the planetary roller screw—the core drive component for knee and hip joints—also suffers from low Production yield, leaving the domestic rate around 20%. The low Production yield and domestic content of these components are acting as bottlenecks to cost reduction.
—What robot development challenges are you currently focused on at Shanghai Jiao Tong University?
First is the manipulation capability of a versatile robot hand (dexterous hand) that fuses vision and touch. We must implement actuators with 27 degrees of freedom within a space the size of a human hand.
Tactile sensing in particular is very difficult. Just as humans can grasp and handle objects with eyes closed, robots must instantly detect pressure, texture, and slip, fuse that with visual information, and carry out complex manipulation.
Second is precise control capability; third is realizing muscle memory and reflexes. Instead of calculating joint angles every time, we focus on studying reflex mechanisms that, by combining the brain's feedback, achieve complex movements and balance with minimal energy and computation.
—Do Humanoid Robot need to closely mimic the human body, such as having five fingers?
Looking at human evolution, the lower body's walking ability may lag behind quadrupeds, but the upper body—especially the versatile "hand" that coordinates with the brain—excelled in manipulation, ultimately making us the rulers of all creatures.
Every object in our world—bottle caps, doorknobs, tools—is thoroughly designed to fit the "human hand." The five-finger structure is the most ideal multipurpose data collector in unstructured, everyday environments. It is also essential for implementing grips that can grasp or handle unexpected objects to cope with diverse variables.
—Some say prices are high and capabilities still fall short of humans, so it is premature to adopt them in real industrial sites.
Whether robots are suitable for industrial sites is judged strictly by calculations based on economic behavior. If a robot replaces the harsh, several-hundred-degree high-temperature environment of a steel mill, adoption is justified even if the device is expensive.
Conversely, for a robot that only tightens screws in a routine factory, there is no need to use top-tier components that can backflip. A uniform standard like how many percent of human ability is needed to break even is meaningless.
Threading a needle is very hard for robots, but in logistics, picking and carrying items can be done far more precisely and efficiently by robots than by people. Rather than absolute price figures or monolithic performance ratings, we must look at the value-to-price ratio required by the specific application scenario.
—How is the replacement of humans by Humanoid Robot in hazardous industrial sites progressing?
The deployment roadmap divides into three stages. First, autonomy begins with simple dangerous tasks. Within 2 to 3 years, a model will be introduced in which humans intervene via remote operation when emergencies or special handling is needed while robots are working.
In 3 to 5 years, robots will judge and perform even complex tasks on their own, evolving into a structure where humans intervene only in highly exceptional cases with few data samples.
—Korea is considered favorable for robot training because manufacturing such as shipbuilding is advanced.
Shipyards are a very promising stage for robots to learn and perform. In the core process of welding, the work environment is so harsh that steel plates get unimaginably hot in summer. And unlike automobiles, ship parts are less standardized, and work must pass through narrow, uneven spaces.
Such irregular and harsh environments are highly suitable for training "embodied AI robots" (embodied AI—robots equipped with AI that can sense, learn, and adapt to their environment like humans). That is why field data from Korean shipyards are so important.
China is also actively attempting to introduce embodied AI robots into welding, grinding, and polishing processes in shipbuilding and repair.
—In what areas do you think Korea is ahead in Humanoid Robot?
Korea has a major strength in reducers, the core of robot joints, especially in harmonic reducers. Since over 10 years ago, it has achieved pioneering innovations, and the kinematic design capabilities currently shown by several universities remain world-class and excellent.