Better Bionics

Mission

Scale up the production of smart prosthetics by implementing continuous self learning.

The challenge

Smart prosthetics are still not available for most amputees. This is caused by various challenges, one of which is recognising which movement a user wants to make. Most prosthetics use muscle signals to control the device. These signals are acquired using a technique called Electromyography (EMG). It has been shown that EMG can be improved using machine learning, however this requires too much power to implement effectively. This is one of the reasons smart prosthetics are still not a common sight, even though they have been in development for a long time, impacting millions of amputees worldwide.

The solution

Other industries have been developing dedicated energy efficient machine learning chips for smartphones and wearables. These are now widely in use to improve for example photography or active noise-cancelling. Computationally, these things are similar to matching muscle data to movements. By implementing these chips in smart prosthetics, it would allow them to learn in real time. This means that a prosthetic will continuously improve its functionality, just by using it. Harnessing the power of these new chips, we will increase the efficiency of manufacturing, providing more amputees with access to smart prosthetics.