Another Shenzhen firm, Lumos Robotics, has deployed robots for tasks including quality inspection and material handling.
Its machines currently work at roughly the same pace as human workers, and getting them to work faster and at greater scale will take time, the company says.
Unlike an artificial intelligence chatbot, which can be trained on vast amounts of text and information available online, a robot needs data from interactions with the physical world, such as how to move, handle objects and respond to its surroundings.
“We need to collect more real-world data. That’s why we’ve developed a number of data-collection tools,” said Yu Chao, founder and CEO of Lumos Robotics.
“The challenge is how to speed up training and iteration … and ultimately produce usable robot skills.”