HANGZHOU, May 16 (Xinhua) -- Inside Zeekr's humming, 5G-enabled electric car factory in the eastern Chinese city of Ningbo, a new type of worker began its apprenticeship.

One robot meticulously sorted components from a shelf, its fingers deftly peeling and applying labels. Nearby, two others coordinated to lift a box from a cart, placing it precisely onto a rack. When one's battery depleted, another autonomously approached to initiate charging.

With a height of an average Chinese man, these UBTECH's Walker S1 humanoid robots offer a glimpse into the future of China's manufacturing sector -- a new wave of automation promising to boost productivity while replenishing the shrinking pool of human workers.

GO TO FACTORIES

Over the past few months, Chinese startup teams have been making waves on the global stage with robots that can perform impressive stunts such as dance routines, backflips and Tai Chi.

Beyond the spotlight, however, some leading robotics firms have been focused on deploying them in factories for more practical jobs. They are joining the global race, led by Tesla Optimus, to integrate humanoid robots into manufacturing.

Shanghai Kepler Robot Co., Ltd. recently released a video of its K2 humanoid robot working at a logistics plant. The robot skillfully navigated the factory floor, handling boxes, transporting goods and operating machinery.

K2 is specifically designed to handle factory work. It has dual arms that can carry 30 kilograms, boasting an impressive eight-hour work cycle on a one-hour charge, said Hu Debo, CEO of Kepler, adding that the base price for its mass-produced version is only 30,000 U.S. dollars.

"If a robot can perform a job as a human does and its cost is around 300,000 to 400,000 yuan (approximately 41,000 to 55,200 U.S. dollars), then it would be cost-effective enough to be deployed," said Xu Jun, head of the innovation technology department at Geely, Zeekr's parent company.

Humanoid robots initially found their application in China's automotive manufacturing sector, driven by the industry's high level of digitalization.

"Automotive manufacturing is one of the most technologically advanced, intelligent, standardized, and data-driven fields in manufacturing, making it an ideal environment for humanoid robots," said Xu.

The robot density has hit 470 units per 10,000 workers in China's manufacturing industry. Over the coming years, the sector is expected to send more intelligent robots to the shop floor.

UBTECH founder Zhou Jian announced that the firm's goal for this year is to manufacture approximately 1,000 humanoid robots, which are set to be deployed in real-world applications to collect more data.

"Application in the manufacturing sector is our priority," Zhou said.

NOT ABOUT REPLACEMENT

China's push for humanoids stems from their potential to bridge the gap left by traditional industrial robots. While industrial robots excel in speed and load-bearing capacity with their pre-programmed, set-path motions, humanoids powered by AI-augmented learning boast greater adaptability.

"Moreover, the large size of industrial robots prevent them from accessing confined spaces like vehicle cabins," explained Xu, adding that humanoids are not intended as replacements of earlier iterations of industrial robots.

Additionally, the "machine-for-human" transition in China's coastal manufacturing plants has proven to be less alarming than initially feared.

"What's really happening in our industry isn't that there are many people lining up to work in factories," said Xu. "The real problem is a labor shortage, especially when production scales up. We simply can't find enough workers."

"Widespread use of humanoid robots could replace humans in hazardous, repetitive, and dull jobs, potentially solving future labor shortages," said Xiong Rong, director of a humanoid robotics innovation center in Zhejiang.

K2 can achieve the same level of output as 1.2 to 2 people in simple and repetitive factory tasks. "Given the labor costs in the Yangtze River Delta, manufacturers can recoup their investment in this robot in just 1.5 to 1.8 years," said Hu.

However, humanoid robots still lag in efficiency for complex tasks.

"Their overall efficiency is about 70 percent of skilled workers' and they cannot perform complex tasks like precision screw-tightening done by senior technicians," said Leng Xiaokun, founder of Leju Robot. The Shenzhen-based firm has trained its robots in several automotive plants to perform box-handling and parts-sorting tasks.

A Shanghai startup has sent its robots to a "technical school". In AgiBot's 4,000-square-meter space, scenes like restaurants, bubble tea shops, and homes are set up.

Over a hundred data collectors, wearing VR glasses and holding controllers, are teaching robots daily chores like folding clothes, clearing dishes, cleaning tables and cashiering in supermarkets. Each action is repeated hundreds of times by the robots.

"Robots have to interact with tangible objects in a 3D world, as such data can't be obtained from the Internet," said Peng Zhihui, AgiBot's co-founder.

Meanwhile, the Beijing-based robotic firm Galbot is exploring an alternative training method: using synthetic simulation data to train robots. The startup has amassed tens of millions of scene data and billions of action data, according to its founder Wang He.

WHY IN CHINA?

China is positioning itself as a powerhouse not just in developing these robots but also in creating an ecosystem for their deployment.

It came as the country has been driving manufacturing digitalization and intelligent transformation, aiming to leverage these technological upgrades to sustain economic growth.

This year's government work report proposed advancing the "AI Plus" initiative to integrate cutting-edge digital technologies with the nation's strong manufacturing base and vast market advantages. It has also planned to develop future industries like embodied intelligence and other next-gen technologies.

At an industrial park in the southern tech hub of Shenzhen, the tightly-knit robotics ecosystem enables seamless collaboration. PaXini Tech supplies tactile sensors to nearby UBTECH, while DexForce streams simulation data directly to AI2Robotics for real-time AI training.

A recent Morgan Stanley report, "Humanoid Robot 100: Mapping the Humanoid Robot Value Chain," has highlighted that Asian companies constitute 73 percent of the top 100 listed firms in this sector, with Chinese firms alone accounting for 56 percent.

China's startups are "benefiting from established supply chains, local adoption opportunities and strong degrees of national government support," according to the report.

Now, cities like Beijing, Shanghai and Shenzhen have established substantial industry funds. In the first quarter of this year, over 50 embodied intelligence firms secured over 6 billion yuan in funding, according to data of IT Juzi, an emerging technology data provider.

A key feature of China's electric vehicle industry is that it has integrated the consumer electronics supply chain, said Li Zexiang, founder of the XBot Park in southern city of Dongguan. "The embodied intelligence industry, exemplified by humanoid robots, is now following suit."

"China has the potential to replicate the disruptive impact from the EV industry in the humanoid space," Reyk Knuhtsen, analyst at SemiAnalysis, told CNBC.

"The influx of humanoid robots into factories will not only boost productivity but also create new industries, giving rise to new industrial chains and job opportunities," said Xu.