Revolutionary South Korean AI Muscle Lifts 4,000 Times Its Weight, Sets New Bar for Robotics

October 31, 2025
Revolutionary South Korean AI Muscle Lifts 4,000 Times Its Weight, Sets New Bar for Robotics
  • A South Korean team has developed a high-performance magnetic composite artificial muscle that uniquely combines softness and stiffness, enabling strong actuation with multidirectional flexibility.

  • The device achieves both high stiffness when needed and substantial stretchability, addressing the typical trade-off in artificial muscles through a tunable stiffness mechanism driven by magnetic components.

  • Researchers see broad applications in soft robotics, wearables, and intuitive human–machine interfaces, with potential to enhance robotics performance across fields.

  • Performance was validated with uniaxial tensile tests to measure strength and elongation to failure, confirming durability for long-term use.

  • In tests, the muscle lifted roughly 4,000 times its own weight, signaling strong potential for future humanoid robotics.

  • The team emphasizes the composite’s ability to combine softness and stiffness to achieve high force and versatile control.

  • Lead author notes the material overcomes the softness–strength trade-off, delivering both high stiffness and large deformability.

  • The artificial muscle can achieve up to 86.4% strain and delivers work density about 1,150 kJ/m³, roughly 30 times higher than human tissue.

  • In its contracted state, the muscle withstands more than double the strain of human muscle, with work density around 30 times higher.

  • The artificial muscle weighs about 1.25 grams and, when stiffened, can bear up to 5 kilograms, roughly 4,400 times its own weight.

  • In the stiff state, the muscle can support approximately 4,000 times its own weight, underscoring its potential for humanoid robots and wearables.

  • Experts note ethical and societal implications of increasingly powerful robotics, reflecting ongoing debates about risks and governance.

Summary based on 2 sources


Get a daily email with more Science stories

More Stories