TIANJIN, May 29 (Xinhua) -- Inside the 1,700-square-meter "Robot Town" at the World Intelligence Expo in north China's Tianjin Municipality, robots were beating drums and playing music.
Also at this venue, a pair of robotic arms, guided by sensors, worked together to thread a needle, while a minimally invasive surgical robot stood nearby and another machine was playing Go. These are just a few examples of how robots are becoming embedded in people's daily lives.
The expo being held from Thursday till Sunday has brought together over 700 exhibitors to showcase cutting-edge technologies, products and application scenarios in the field of artificial intelligence (AI).
"At this expo, we can see the integration of AI and robotics across many fields, such as healthcare and manufacturing. My feeling is that China is leading the industry in robotics and making rapid progress in AI," Andrea Rösinger, representative from a German technology company, told Xinhua.
Statistics show that China's embodied intelligence industry is leaping forward with an annual growth rate of over 50 percent. According to a report released by the Development Research Center of the State Council, China's embodied AI market is expected to surpass one trillion yuan (about 146.54 billion U.S. dollars) by 2035.
ROAD TO SELF-RELIANCE
The development of embodied AI requires, on the software side, more powerful algorithms and large models. On the hardware side, it places extremely high demands on materials, chips and high-end sensors.
Yu Tong, international sales manager of Galileo (Tianjin) Technology Co., Ltd., was busy briefing a German merchant group about the quadruped robots in his company's booth.
When the company was founded, core technologies were monopolized by foreign firms, and there were no mature products in China that they could learn from. The team initially assembled parts from different suppliers, but the resulting leg structure was too bulky for flexible and precise movement.
So the engineering team decided to start from scratch. They developed their own motors, reducers and explosion-proof casings, and integrated core components into a modular joint system. The integrated sensors can precisely measure torque, position and velocity in every step, feeding real-time data back to gait algorithms. The algorithm then calculates, in milliseconds, the force and angle required for the next stride.
This closed-loop mechanism is what makes their robots intelligent. When climbing a 35-degree slope, for example, the robot does not simply charge ahead. Instead, it chooses a more stable method -- it first steps backward and then climbs up the slope.
These independent R&D efforts have certainly paid off for the company. Adapting to stairs, pipelines, sandy ground and other complex conditions, their robots have already been deployed on a large scale in rail transit, power grids and explosion-proof petrochemical facilities. Galileo has now accumulated more than 300 patents and software copyrights ranging from integrated joint modules to locomotion algorithms.
Beyond participation at the expo, Galileo has opted to base its major production lines in Tianjin. "We make use of Tianjin's precision manufacturing to realize mass production or customized orders quickly, and nearby talent and research resources in Beijing, as well as application scenarios in Hebei Province, are also handy for us," Yu said.
"The rapid development of embodied AI would not be possible without China's solid, full-chain manufacturing base. Meanwhile, China's traditional manufacturers and emerging industries, which have a huge appetite for better, more automated machinery, provide our next-generation embodied AI industry with a vast market," Yu explained.
OPEN INNOVATION ECOSYSTEM
China's manufacturing base underpins the hardware side of embodied AI, while the software side is growing through an open, collaborative ecosystem developed by Chinese companies and research institutions.
During the expo, PaXini Technology has released its ten-billion-scale multimodal embodied AI dataset and cross-border embodied AI data case.
Just as AI needs large amounts of quality data for training, embodied AI requires real-world interaction data to learn how to complete tasks autonomously in complex environments.
There is a severe shortage in this regard. Embodied AI needs hundreds of petabytes of physical interaction data, and recent industry data reveals a huge gap.
In a bid to reduce this shortfall, PaXini, a haptic technology and humanoid robotics company, has built five super data-collection factories across China. Inside these factories, technicians wear self-developed tactile gloves equipped with 30 six-dimensional tactile modules and perform real-world operations like grasping, assembling and sorting. The data is collected across 15 sectors, including auto manufacturing, healthcare and retail.
PaXini has commercialized its data capabilities by launching a data cloud mall with industrial partners. Users can purchase standardized, high-quality datasets on demand and use them directly for model training without collecting data by themselves, thereby sharply reducing both the time and cost of data acquisition, said Xu Jincheng, founder and CEO of PaXini.
PaXini's training data service is also open to international clients. During the ongoing expo, the company has completed a landmark cross-border transaction involving embodied AI training data, selling it to an overseas large model company under Tianjin's pioneering negative list policy for data outbound management.
Such openness is not unique. In a similar spirit of collaboration, Chinese mapping service provider AutoNavi has fully open-sourced its ABot-M0 platform, including robot datasets, model architecture, training framework and tool chains.
This month, China launched a national-level artificial intelligence pilot-testing base dedicated to embodied intelligence in the eastern city of Hangzhou, where companies can rent services such as computing power, data access, model support and scenario validation.
"China's AI industry has always pursued an open, co-creative ecosystem, which lowers the threshold for innovation and reduces duplicated R&D costs across the industry, greatly stimulating innovation vitality," said Wu Qingxiang, associate professor at the College of Artificial Intelligence, Nankai University in Tianjin.
In recent years, China's AI innovation has grown rapidly. China has become the world's largest holder of AI patents, accounting for 60 percent of the global total, while its core AI industry has exceeded 1.2 trillion yuan in scale, according to the National Data Administration.
Nobel laureate economist Michael Spence recently hailed China's emphasis on the adoption and application of AI in its economy, which he considered something not all countries can achieve.
"Chinese companies can offer overseas users cost-effective and reliable hardware-software solutions, while research teams are well-positioned to tackle common technical challenges with global partners. This will accelerate embodied AI adoption worldwide and drive the industry toward a more coordinated, advanced paradigm," Wu added. ■
