On December 2, 2025, ByteDance, based in the Chinese mainland, unveiled the world’s first robot capable of tying shoelaces, marking a leap in robot dexterity beyond lab settings. Powered by a custom reinforcement learning model, the robot achieved an 83% success rate in threading and knotting shoelaces—nearly double the 45% rate of robots trained through traditional imitation methods.
This advancement showcases how reinforcement learning, which lets machines learn from trial and error, can handle longer and more delicate tasks. For young innovators, it underlines the potential for AI-driven automation in sectors from manufacturing to healthcare, where fine motor skills are crucial.
Looking ahead, researchers plan to apply this model to other intricate tasks such as assembling electronics or performing complex surgical assistance. As robots move from controlled labs into real-world environments, breakthroughs like this bring us closer to a future where machines seamlessly blend into daily life.
Reference(s):
cgtn.com
