The robot showcased its ability to thread a needle and stitch, handling delicate, flexible materials with sub-millimeter precision during a live demonstration on Dec. 22. Hand embroidery, while niche, is considered one of the most complex tasks in robotics, requiring precise vision, adaptive force control, and the coordination of both hands while managing constantly shifting materials, according to Interesting Engineering.
The humanoid robot performed the task smoothly, maintaining stability throughout. This marks a major milestone for robotics, as such intricate manual tasks were once considered a longstanding challenge beyond the capabilities of automation.
China's TARS Robotics humanoid robot performs hand embroidery. Video courtesy of TARS Robotics
Dr. Chen Yilun, CEO of TARS Robotics, explained that the success comes from the company's DATA AI PHYSICS trinity approach, which integrates real-world data, AI models, and physical robotic systems. He highlighted that this system minimizes the gap between digital models and real-world execution, enabling reliable performance. The approach also supports scalable development and aligns with the Scaling Law for AI systems.
Founded in February and based in Shanghai, TARS is led by a team of veterans from Baidu's Apollo project and Huawei’s autonomous driving division, TechinAsia reported.
TARS' success with embroidery highlights the robot's potential to be adapted for more complex industrial tasks. Humanoid robots are increasingly being adopted in industrial settings. STMicroelectronics, a European chipmaker, recently announced a partnership with Italy’s Oversonic Robotics to deploy its humanoid robot, RoBee, at global production sites, marking the first use of humanoid robots in semiconductor manufacturing.
Meanwhile, CATL, the world's largest electric vehicle battery manufacturer, has begun deploying humanoid robots at its Zhongzhou facility in central China, becoming the first to use humanoid machines extensively for industrial automation.