A new human-machine interface called the myokinetic control interface is proposed for controlling hand prostheses using the movement of multiple magnets placed in residual limb muscles.
Machine learning models, such as linear and radial basis functions neural networks, are used to improve the localization of these magnets and translate magnetic data into commands for prosthetic devices.
The system achieved high tracking accuracy (720 μm) and low latency (12.07 μs) in a test environment, and it is designed to be more power-efficient than previous methods, paving the way for future research on managing multiple magnets at once.