Abstract Vocabulary as Base for Training with Pattern Recognition EMG Control

The uprising of multi-channel wearable EMG sensors combined with machine learning pattern recognition algorithms offers the possibility to control multiple degree of freedom hand prosthetics. Such human-machine interaction systems require training from the user, mostly to link gestures with underlying EMG patterns. As intended end users have a missing hand, the question arises how to train them to use myo-electric prosthetics without instructing them to perform gestures; A key element to start training with pattern recognition EMG based prosthetic control is creating a shared vocabulary with the participant/patient. The shared vocabulary forms the base for the explanation and communication about the pattern recognition EMG. In this research an abstract form of communication based on animal sounds is used to form a shared vocabulary for a child with missing hands. We found that the abstract communication worked well and motivating when explaining pattern recognition EMG to a child. The communication tool that gives additional interaction makes the explanation much clearer since the participant starts directly with experiencing the pattern recognition EMG. Also, it is concluded that the abstract nature of the tested communication allows the participant to keep an open mind for gestures other than normal healthy hand movements when exploring the possible control contractions. Thus, abstract based communication can offer benefits during the training with pattern recognition EMG.