Robotics Competition Highlights China’s Push to Bring Embodied AI Into Real-World Applications

The development of embodied artificial intelligence is moving beyond laboratory demonstrations as robotics researchers and technology companies increasingly focus on machines that can understand physical environments and perform practical tasks.

A national robotics competition held in Chongqing has highlighted this transition, bringing together more than 100 teams working on technologies ranging from biomimetic robots and industrial automation to AI systems designed to perceive and interact with the physical world. The event was held in Shapingba District, where local authorities are seeking to strengthen the region’s position in the emerging embodied-AI industry.

The development is significant because embodied AI represents a different direction from conventional generative AI. Instead of operating primarily through text, images or digital information, embodied AI connects intelligence with physical machines that must sense their surroundings, make decisions and carry out actions.

From Digital AI to Physical Intelligence

Generative AI has largely developed through systems capable of processing and producing digital information.

Large language models can write text, generate software, analyse documents and answer questions. Robotics introduces an additional layer of complexity because an intelligent machine must interact with the physical world.

A robot operating in a warehouse, hospital, factory or home cannot rely only on a written instruction. It must understand where objects are located, determine how they can be handled and respond when conditions change.

This combination of perception, reasoning and physical action is at the heart of embodied AI.

More Than 100 Teams Take Part

The Chongqing competition brought together more than 100 teams from across China.

The event was the final stage of the 11th “Maker in China” SME Innovation and Entrepreneurship Contest for Intelligent Biomimetic Robots. The three-day national competition concluded on September 11 in Shapingba District.

Participants included research institutions, university teams and technology companies.

The projects covered multiple areas, showing that embodied AI is not limited to humanoid robots. Developers are working on industrial systems, assistive machines, inspection robots, autonomous drones and other specialized platforms.

Real-World Deployment Is Becoming the Main Focus

One of the most important developments highlighted by the competition was the emphasis on practical deployment.

Rather than demonstrating robots only under controlled laboratory conditions, participating teams were encouraged to address real operational requirements.

Projects included systems designed for the electric-power sector, healthcare and service industries. Other developments focused on warehouse coordination, rehabilitation assistance and infrastructure inspection.

This represents an important shift in robotics development.

A machine may perform impressively in a controlled demonstration, but commercial deployment requires much greater reliability. It must continue operating when objects move unexpectedly, lighting changes, instructions vary or the environment becomes crowded.

Robots Need to Understand Their Environment

Embodied AI depends heavily on environmental perception.

A robot needs sensors and AI systems that allow it to identify objects, understand spatial relationships and detect changes around it.

For example, a warehouse robot may need to distinguish between different packages while simultaneously avoiding other machines and workers.

A healthcare robot may have to operate around people whose movements cannot always be predicted.

These challenges require AI systems to combine visual information, physical sensing and decision-making.

The Importance of Multi-Robot Coordination

One area showcased during the competition involved coordination between multiple robots.

In large industrial environments, a single robot may not be sufficient to complete an entire workflow.

Several machines may need to divide tasks, communicate with one another and determine an efficient sequence of operations.

AI-based task scheduling can potentially allow robots to coordinate their activities dynamically.

This could become particularly valuable in warehouses and manufacturing facilities where large numbers of machines operate simultaneously.

Assistive Robots Could Support Healthcare

Another area of development is assistive robotics.

Competition projects included exoskeleton technologies intended to support older adults and rehabilitation applications.

Such systems could potentially assist people with mobility challenges by providing physical support during movement.

However, healthcare robotics also requires extremely high safety standards.

A machine operating close to a person must respond appropriately to unexpected movements and must not apply excessive force.

This makes reliable perception and control essential.

Robots for Infrastructure Inspection

Some teams focused on biomimetic robots capable of inspecting electric-power infrastructure.

Infrastructure inspection can involve environments that are difficult or dangerous for humans to access.

Robots could potentially perform repetitive inspection tasks while collecting visual or sensor-based information.

AI could then help identify abnormalities or prioritize areas requiring human attention.

This represents one of the practical applications where robotics and artificial intelligence can complement human workers rather than simply replacing them.

Autonomous Clothes-Folding Robot

Among the projects highlighted in the competition was an autonomous clothes-folding robot.

Although folding clothing may appear simple from a human perspective, it is technically challenging for robots.

Clothing is flexible rather than rigid. Its shape can change significantly depending on how it is placed, and the robot must determine how to grasp and manipulate it.

A successful system therefore requires accurate perception and precise physical control.

The project illustrates how apparently ordinary household activities can become demanding engineering problems for autonomous machines.

Biomimetic Robotics

The competition’s focus on intelligent biomimetic robots also reflects a broader trend in robotics research.

Biomimetic systems attempt to draw inspiration from biological structures and movements.

Examples can include robots inspired by animals, insects or human biomechanics.

Such designs may help researchers develop machines capable of moving through difficult environments or performing specialized tasks.

The goal is not necessarily to reproduce nature exactly, but to learn from biological mechanisms when designing robotic systems.

Small Flying Robots and New Designs

Another project highlighted by the competition involved a miniature flapping-wing drone inspired by hummingbirds.

Small flying systems could potentially be useful for inspection, monitoring and other specialized applications where conventional aircraft or larger drones may not be practical.

Miniaturization, however, introduces its own engineering challenges, including energy efficiency, flight stability and control.

AI can play a role in helping such machines interpret their environment and adjust their behaviour.

The Hardware-Software Connection

Embodied AI requires much closer integration between software and hardware than traditional digital AI.

A language model can produce an answer without physically interacting with its surroundings.

A robot must translate an AI decision into movement.

That means the system needs mechanisms for controlling motors, joints, sensors and other physical components.

A mistake in software can therefore have a direct physical consequence.

This is why robotics developers must combine AI research with mechanical engineering, electronics, control systems and safety engineering.

Why Competition Matters

Technology competitions can provide an environment where researchers and companies test ideas under common conditions.

They also create opportunities for universities, startups and established companies to interact.

The Chongqing event included discussions involving companies, universities and research institutions, with activities related to technology transfer, financing and commercial deployment.

This connection between research and industry is important because promising laboratory technologies often require additional engineering and investment before they can become commercial products.

Shapingba’s Ambition in Embodied AI

The competition is also connected to Shapingba District’s broader effort to develop an embodied-AI industry.

Local authorities have been attempting to use the district’s existing industrial and research infrastructure to attract robotics companies and encourage technology commercialization.

According to local government data cited in the event coverage, the output value of the district’s robotics industry increased by 233 percent year on year during the first seven months of 2026.

The figure reflects local industrial growth, although it should not be interpreted as a measure of the performance of the entire Chinese robotics sector.

The Commercialization Challenge

Building a robot that works in a demonstration is only the beginning.

For widespread commercial adoption, machines need to be affordable, reliable and easy to maintain.

They must also operate safely around people and integrate with existing industrial systems.

This creates a significant gap between technological demonstrations and mass deployment.

Embodied AI developers are therefore increasingly focusing on practical performance rather than demonstrations alone.

AI Must Deal With the Physical World

The physical environment is far less predictable than a digital interface.

Objects can fall, people can move unexpectedly, surfaces can change and machines can encounter situations that were not included in their training data.

An embodied AI system must therefore continuously observe its surroundings and adjust its actions.

This creates a feedback loop involving perception, reasoning, action and feedback.

The ability to operate effectively within this loop is one of the defining challenges of modern robotics.

Implications for Industry

If embodied AI technologies become sufficiently reliable, they could influence several industries.

Manufacturing could use robots for flexible production.

Warehouses could use autonomous machines for sorting and transportation.

Healthcare could adopt assistive robotic systems.

Energy companies could deploy robots for infrastructure inspection.

Service industries could use robots for selected repetitive tasks.

These applications could increase automation while also changing the types of skills required from human workers.

Human-Robot Collaboration

The future of robotics is not necessarily limited to fully autonomous machines.

In many environments, robots may work alongside humans.

A robot could perform physically demanding or repetitive activities while people handle communication, judgment and tasks requiring flexibility.

This model of human-robot collaboration may prove particularly useful in workplaces where complete automation is technically difficult or economically inefficient.

Safety Will Remain Critical

As robots become more intelligent and autonomous, safety becomes increasingly important.

A machine capable of making decisions and physically interacting with its environment must have safeguards that prevent dangerous behaviour.

Developers need to test how robots respond to unexpected situations, equipment failures and changes in their surroundings.

Safety considerations will become especially important as robots move from factories into hospitals, public spaces and homes.

The Next Stage of AI Development

The Chongqing competition demonstrates how the AI industry is gradually expanding from digital intelligence toward physical intelligence.

The next generation of AI systems may not simply answer questions or generate content. They may also interact with objects, operate machines and perform tasks in the physical world.

That transition could significantly expand the practical applications of artificial intelligence.

But it will also require advances in hardware, software, safety systems and AI reasoning.

Conclusion

The robotics competition in Chongqing provides a snapshot of the growing effort to turn embodied AI from an experimental technology into practical industrial systems.

More than 100 teams participated, with projects covering robotics, industrial automation, healthcare assistance, infrastructure inspection, warehouse coordination and other applications.

The most important development is the increasing emphasis on real-world performance.

For embodied AI to become a major part of the technology economy, robots will need to do more than demonstrate impressive movements. They will have to understand changing environments, perform useful tasks reliably and operate safely alongside people.

The competition therefore reflects a broader transformation in artificial intelligence: the move from AI that primarily works with information toward AI that can increasingly perceive, decide and act in the physical world.