Humanoid robots complete a 24-hour supermarket challenge.

Six humanoid robots faced an unusual challenge: autonomously managing an operational supermarket in Shanghai for 24 consecutive hours. The test, which concluded on November 3rd, was organized by Zhiyuan (智元) as the final event of its AI WAVE series. Of the six participating robots, four successfully completed the 'work challenge,' virtually earning themselves a 'permanent position.'

The experiment represents a significant test for "embodied AI," which studies how to equip robots with a physical body to interact with the real world. For years, humanoid robots have often been criticized for being more suited to spectacular demonstrations, such as performing flips or tai chi, than to carrying out useful tasks. This practical test in a real commercial environment aims to prove the opposite.

The idea of robots performing human jobs is not new, but for decades it has faced enormous technical difficulties. Industrial robots, fixed and programmed for repetitive tasks, have been common since the 20th century, but creating versatile and autonomous humanoid robots is a much more complex challenge. The concept of "embodied AI" is gaining momentum in the 21st century with the advancement of artificial intelligence: instead of being just a brain in a box, AI must learn through a body that perceives and acts in the environment.

The industry has gone through a phase of spectacular demonstrations to attract investment, but the gap between these and practical, reliable application has remained wide. Starting in 2022, some supermarket chains began introducing robots for specific tasks, but entrusting an entire store to robotic management was considered impossible. The Zhìyuán 24-hour challenge marks a concrete attempt to bridge this gap, testing not only the robots' physical abilities but also their capacity to make decisions, adapt to unexpected situations, and work continuously. The partial success (4 out of 6 robots) indicates progress, but also the long road still ahead for full automation.