Artificial Intelligence in Upper Limb Prosthetics
Artificial Intelligence (AI) is fast becoming an integral tool in the business world. We’ve all probably used AI in some form daily and don’t even realise it. One of the most common types of AI is machine learning and we are now seeing it used in developing the next generation of prosthetic hands.
Modern prosthetic hands can now be operated with multiple dexterities. A big step forward from even the best prosthetic hands 20 years ago in terms of degrees of freedom and control systems. However, there are no commercially available prosthetic hands that can provide satisfactory sensory feedback to the user.
Things are about to change. Researchers from Florida Atlantic University's College of Engineering and Computer Science are creating prosthetic hands that can “feel” by incorporating liquid metal tactile sensors on the fingertips and using machine learning algorithms to differentiate surface textures and how fast the hands moved over the surfaces. (Video link) Furthermore, encapsulating the liquid metal sensors within silicon-based elastomers the researchers were able to improve conductivity, compliance, flexibility and stretchability.
The team trained 4 machine learning algorithms to differentiate various speeds when fingertips of a prosthetic hand slide across different surface textures at various speeds. The results revealed that the liquid metal sensors and algorithms were able to accurately differentiate all 10 surfaces and various speeds of which the fingertips moved over the surfaces.
There is of course more development required before we develop commercially available prosthetic hands that can truly “feel” our environment, but this brings us a step closer to the next level.
