Skip to content

Research

Humanoid robots test delivery, stocking and checkout in Daegu

Edited and reviewed by AI Ping Editorial Team

3-line summary

  1. Humanoids took on delivery, shelf stocking and checkout in the International Robot Contest’s first supermarket challenge in Daegu.
  2. The three jobs appeared in both virtual and physical settings. The Nov. 14 schedule grouped delivery with stocking and put checkout in a separate column.
  3. The organizers scheduled judging and an awards session for the challenge. Public notices showed no team-by-team scores or rankings.
BeginnerBeginnersDevelopers
A white humanoid robot holds a box of goods beside shelves in a fictional supermarket aisle.
AI-generated conceptual image · A fictional supermarket aisle and humanoid illustrate the delivery task. · AI Ping

Humanoid robots competed at a mock supermarket in Daegu, South Korea, on Nov. 14, taking on three familiar store jobs: moving goods, stocking shelves and handling products at checkout. The International Robot Contest added the supermarket challenge this year. Its organizers assigned each task in both a computer simulation and a physical test environment, giving teams two settings in which to show what their robots could do.

South Korea’s Ministry of Trade, Industry and Resources and the city of Daegu hosted the two-day contest at EXCO. The Korea Institute for Robot Industry Advancement organized the event, and Soongsil University and Kairoba ran the supermarket challenge. Rehearsals took place on Nov. 13. News1’s photographers recorded a robot moving toward shelves and a separate checkout scene that day. South Korea’s Electronic Times reported on Nov. 15 that the competition tested all three tasks in simulated and physical store environments.

Delivery and stocking shared a slot, with checkout separate

The challenge guide labeled delivery as task A, stocking as B and checkout as C. The labels correspond to moving goods, placing them on shelves and processing them at the counter. The guide lists a simulated and a physical version of each job. That setup put the digital exercise and the work of handling objects with a robot in the same competition.

The official schedule listed the three jobs in separate columns for Friday’s rehearsals. For Saturday’s competition, it grouped delivery and stocking in one column and left checkout in a second. Six teams appeared on that timetable: Workerholic, SUPERB AI, SHAPE, ee-ai-lab, Robotguk and CILAB. The 2:30 p.m. slot listed SHAPE for delivery and stocking and CILAB for checkout. Both entries were marked for live coverage. In the first morning slot, the schedule paired SUPERB AI on delivery and stocking with Workerholic at checkout. The separate columns put different teams at the two stations in a given slot.

Delivery, stocking and checkout were each assigned in both a virtual store and a physical test area.
The supermarket challenge assigned each of its three tasks in simulation and on a physical test floor. · Source International Robot Contest · Graphic by AI Ping

A store task moves from a screen to a physical space

A simulated pickup represents an action on screen. In a physical run, a robot has to make contact with an object and move it to another location. A checkout task, for example, can include bringing a product over to a counter. A mock aisle makes it possible to observe the robot moving through space as well as handling the item. The timetable does not report whether teams reused their simulation programs unchanged or how much help people gave individual robots.

A dark robotic gripper carries a pale green carton over a fictional checkout counter.
AI-generated conceptual image · A fictional gripper, carton and counter illustrate handling goods at checkout. · AI Ping

The Seoul National University robotics club SHAPE lists an October qualifying submission period and the Nov. 13–14 final at EXCO in its project calendar. The event timetable also names the club in rehearsal and competition slots. SHAPE describes recognizing objects, navigating and manipulating objects as skills that carry across contests. Its public schedule also lists a separate application for a home service robot competition. For robotics teams and retailers, the published program shows which store tasks the event put in both simulated and physical settings. It gives no evidence about how the robots would perform in a busy store over a full shift.

The timetable listed judging and a separate awards session

The supermarket schedule set judging for 5:30 p.m. on Saturday and its own awards session for 6 to 6:30 p.m. The Electronic Times, reporting the organizers’ account, said about 4,000 competitors and spectators visited the wider contest over two days. That figure covers the full event. The supermarket challenge’s official page lists 40 participants across its two days, breaking the figure down as 21 on the first day and 19 on the second.

The same post-event account reported 136 awards across the overall contest, including four presidential awards. One went to a team in the intelligent humanoid division of the separate Intelligent Robot Competition. The main event schedule lists that division and the supermarket challenge as different competitions. As of Nov. 15, the supermarket challenge page and the event’s public notices showed no team-by-team scores or final rankings for delivery, stocking and checkout.

Sources