Artificial muscles have learnt to "think" and memorise movements

Researchers at the Georgia Institute of Technology are creating technologies that are closer to humans not only in form but also in feel. Their designs are changing the way robotics is viewed, offering soft and adaptive solutions inspired by biology.

Professor Hong Yeo from the Georgia School of Mechanical Engineering is working on artificial muscles that mimic the structure and behaviour of real tissue. His team uses layered fibres similar to tendons and muscles, and trains AI to control them in real time.

These muscles don't just follow commands - they memorise movements, learn and adapt, providing the natural fluidity of movement so essential when rehabilitating after injury or limb loss.

"Such movements are not just about functionality. It's a returned independence, confidence and sense of self," explains Yeo.

One of the first practical realisations was a prosthetic glove based on soft artificial muscles (results published in ACS Nano). Unlike traditional rigid structures, the new glove moves in sync with the user's intentions, senses grip strength, suppresses tremors and adapts to movements.

Soft fibres inside the glove bend, stretch and twist to help perform subtle actions such as buttoning a button, lifting a cup or holding a child's hand.

"It's not just movement - it's freedom returned," Yeo says.

Yeo's journey into biomedical engineering began after the sudden death of his father when he was still at university. He originally dreamed of designing cars, but tragedy changed his priorities.

"I thought: maybe I can do something that actually saves lives," he recalls.

Since then, his goal has been to create technologies that help people regain what they've lost.

Creating such "living" tissues is a challenge. The materials must be both flexible, strong and safe, not cause an immune response and be easily integrated into the body.

Yeo and his team achieve this by calibrating each fibre as a precision instrument, achieving a muscle memory effect where the material becomes accustomed to movements and adapts to changes.

"If it becomes part of the body, it has to work with it, not against it," the scientist emphasises.

Yeo's work requires the co-operation of engineers, biomedical scientists, material scientists and programmers. This is the only way to create truly effective and safe devices.

His laboratory is actively collaborating with clinics and industrial partners to transfer developments from academia to real medical practice.

The future of robotics, according to Yeo, will be defined not by power, but by feel.

"If a device feels foreign - it won't be used. But if it feels like a part of the body - it has the potential to change lives," he explains.

This is the opposite of the Terminator image. Here, machines don't replace humans, but help them reclaim themselves.

Read the original research here:

• Saewoong Oh et al, Empowering artificial muscles with intelligence: recent advances in materials, designs, and manufacturing, Materials Horizons (2025). DOI: 10.1039/d5mh00236b
• Tae Woog Kang et al, Soft Nanomembrane Sensor-Enabled Wearable Multimodal Sensing and Feedback System for Upper-Limb Sensory Impairment Assistance, ACS Nano (2025). DOI: 10.1021/acsnano.4c15530