SerpApi.Org
A humanoid robot demonstrated its cooking skills by preparing traditional Xinjiang dishes during a live culinary show in China.
The robot grilled kebabs and skewered meat, carrying out tasks that required precise hand, wrist, and arm coordination as well as careful force control when handling soft, uneven ingredients.
After less than a week of specialized training, the machine completed basic cooking operations, although its performance remained well below that of experienced chefs.
Recently, UniX AI unveiled its Panther home robot, an AI-powered wheeled humanoid that can cook, clean, organize, and assist users with 8–16 hours of runtime.
Robot kitchen skills
A humanoid robot took on one of the most challenging tasks in robotics—preparing food—during a live cooking demonstration in northwest China’s Xinjiang Uygur Autonomous Region, showcasing the latest advances in embodied artificial intelligence and robotic dexterity.
The demonstration was held on July 15 at Liuxing Street in Yining City, a national-level tourism and leisure district known for its distinctive hexagonal street layout. Visitors gathered at a local restaurant to watch the robot prepare traditional Xinjiang specialties, highlighting how humanoid robots are beginning to tackle complex real-world tasks that demand precision, adaptability, and delicate manipulation, reports Orient.
The robot first grilled kebabs and skewered meat, tasks that required precise coordination between its fingers, wrists, and arms while carefully controlling the amount of force applied to soft, irregular pieces of meat. According to the developers, the machine learned these basic operations after less than a week of task-specific training. Although its performance remained well below that of an experienced chef, the demonstration showed significant progress in robotic manipulation.
The greatest challenge came with hand-pulled noodles, one of Xinjiang’s signature dishes. Stretching the soft, elastic dough requires subtle adjustments in pressure and movement to avoid tearing or dropping it—something that even novice human cooks find difficult.
AI tackles cuisine
To tackle this task, the robot is equipped with tactile sensors embedded in its dexterous hands. These sensors continuously measure pressure, movement, and contact forces as the dough is stretched. The system records force data from multiple directions during every attempt, enabling engineers to analyze each movement and refine the robot’s performance through repeated training and machine learning.
Even though it is not specified which robot model was used, developers said every cooking session generates valuable real-world data that helps improve the robot’s ability to manipulate delicate objects and perform increasingly complex tasks. Unlike controlled laboratory experiments, preparing food exposes robots to constantly changing conditions, making it an effective test of embodied AI systems that must combine perception, motion planning, and force control in real time, reports Newsflare.
The demonstration also highlighted the growing role of practical field testing in advancing humanoid robotics. Engineers believe exposing robots to everyday activities such as cooking will accelerate their development for applications beyond manufacturing, including hospitality, retail, healthcare, and domestic assistance.
The cooking show served as a preview of the second World Humanoid Robot Games, which will take place in Beijing from August 22 to 26. The event is expected to showcase the latest achievements in humanoid robotics through competitions based on real-world tasks, providing another benchmark for measuring how quickly robots are progressing from controlled demonstrations toward practical everyday use.
