At the World Artificial Intelligence Conference (WAIC) in Shanghai, industry insiders said Chinese robotics companies face major hurdles in improving how their robots interact with the physical world. One key problem is a lack of sufficient data. Another is the need for a better 'brain' – the artificial intelligence that controls the robot's actions. Wang Xiaogang, co-founder of SenseTime and chairman of its robotics division, explained that the biggest challenge for embodied AI – which combines hardware, software, and real-world interaction – is creating a closed-loop system. This means linking hardware, data, models, and real-world scenarios so they can continuously improve each other. Without a good 'brain' and enough data, robots cannot learn to handle everyday tasks effectively. For example, a robot might struggle to pick up an object it has never seen before or navigate a cluttered room. The industry is working to collect more data and develop better AI models, but progress is slow. Experts at the conference emphasized that solving these issues is critical for China to advance in robotics. The country has strong hardware manufacturing, but software and data remain weak points. Companies are now focusing on gathering high-quality data and improving AI algorithms to make robots smarter and more useful in homes, factories, and other settings.