China's advancements in humanoid robotics are helping to bring robots out of science fiction movies and into the real world.

The "Robot Gala" was recently staged by major Chinese robot maker AgiBot in Shanghai, featuring an all-robot cast. It was a showcase of modern robots’ motion control, computer vision, and creativity.

"Usually, robots are just there to support the show. This time, we want them to be the stars," said Li Jiayu, Director of Solutions at AgiBot.

Last year, eight out of ten humanoid robots sold were from Chinese vendors. Over 140 robot makers in the country competed to create the best products, according to the Ministry of Industry and Information Technology.

Robots can also move from the dance floor to the factory floor. Chinese engineer Chang Liu is building software in Reading, UK that allows users to train robots. They learn from every interaction. And with enough practice, they could eventually be ready for the production line, helping companies work faster and smarter.

"Robotics is the opportunity that would bridge the labour gaps. That would ultimately produce enough productivity that would serve the need for future societies," said Liu, CEO, Extend Robotics.

Humanoid robots look set to perform more and more tasks once reserved for humans, including making food, delivering parcels, and even working in hospitals.

In a store in the heart of Shanghai, a robot worker is busy welcoming guests.

"It answers customer questions, interacts in the store, and helps create a lively atmosphere," said Zhang Qian, the store manager.

AgiBot recently took part in a tech show in London where it introduced robots that can help students and teachers in the classroom.

"In China, we're already working with a secondary school already. We built up some courses and lessons for the robots to work with the students," said Martin Ma, AgiBOT UK Country Manager.

Training robots takes time and money. But Elliot Wu, an assistant professor in Machine Intelligence at the University of Cambridge, says that is changing fast. He's working on software that would allow robots to learn simply by uploading a video.

"Potentially that's going to be a hundred or even a thousand times faster in terms of the scale of the data that we can collect in a certain period of time," said Wu.