AGIBOT G2 Robots Complete 1,730 Tasks in 24-Hour Shanghai Test

Six AGIBOT G2 humanoid robots were tested in a 24-hour retail challenge at a convenience store in Shanghai’s Xuhui district from November 2 to 3, 2026. They completed 1,730 tasks. Two robots failed at their duties of cleaning and shelf restocking.

A Chinese robotics company livestreamed an experiment to test its humanoid machines in a real-world retail environment. The Robots Never Clock Out challenge placed six AGIBOT G2 robots in a functioning convenience store to work alongside human staff and customers for 24 hours. Reports came from NDTV and Tech Times.

Robots Complete 1,730 Tasks, but Two Fail

The livestreamed event ran from 10am on November 2 to 10am on November 3. The robots were assigned various retail duties. Over the 24-hour period, the six machines collectively completed 1,730 tasks. These included 683 customer greetings, 369 item retrievals and deliveries, 147 coffees prepared, 126 hot-food orders handled, 277 shelf inspections and restocking jobs, and 128 table-cleaning tasks. This data was detailed by NDTV and Maharashtra Times, according to Maharashtratimes.

The company divided the challenge into four operating periods. These were the lunch rush, the evening rush, late-night duty, and breakfast preparation. The event placed the robots in a shared space alongside actual customers and store staff. This is not a controlled setting where engineers could reset conditions between attempts. This combination of demands is meaningfully harder to manage than a single, repeatable laboratory task.

Humanoid Robots Still Struggle with Hardware Control

While the robots completed a high number of tasks, Tech Times noted a key detail. The reviewed materials do not provide an independently audited task log, intervention count, or failure rate from the event.

AGIBOT G2 Robots Complete 1,730 Tasks in 24-Hour Shanghai Test
Photo: ndtv.com

A Test of Embodied AI in a Real World

The Robots Never Clock Out event marks a shift. It moves from short laboratory demonstrations to extended testing in a public, semi-structured environment. Navigation, shelf manipulation, customer interaction, and task-switching all unfold at once in an aisle shared with real people.

AGIBOT tasked the G2 units with stocking shelves, preparing coffee and hot food, greeting customers, and clearing rubbish. The company divided the challenge into four distinct periods: lunchtime rush, evening rush, late-night duty, and breakfast preparation.

The experiment was designed to test how the machines perform under different working conditions, not in a controlled scenario. The mixed performance suggests that humanoid robots still have considerable progress to make. They must reliably handle the full range of responsibilities expected of human workers. Until these machines are as reliable, productive and cost-effective as people, replacing human shop workers may prove considerably harder than simply putting a robot behind the counter. Tech Times concluded this.

Photo of author

Sophie Lin - Technology Editor

Sophie is a tech innovator and acclaimed tech writer recognized by the Online News Association. She translates the fast-paced world of technology, AI, and digital trends into compelling stories for readers of all backgrounds.

Stanford Medicine Researchers Develop New Tool to Predict Alzheimer’s

Honor 700 Series: Rear Screen & 200MP Camera Leaked Ahead of November Launch