Minimal Forcemyography Human-Machine Interface for Hand Gesture Recognition

This study proposes a very simple Human-Machine Interface (HMI) to recognize six basic hand gestures. The device is equipped with only two forcemyography (FMG) sensors applied to specific forearm muscles and it is a scaled-down version of a previous one used for gesture recognition and control of exoskeletons and prostheses. Data collected from five healthy subjects were considered. A double-threshold segmentation technique of the two FMG signals allowed the extraction of features (i.e., mean, standard deviation, root mean square, area, slope) to recognize the hand gestures. Features were recalculated every 100 ms to provide real-time performance. Classification performance was evaluated using both Linear Discriminant Analysis (LDA) and a Linear-Support Vector Machine (L-SVM) classifier and applying 10-fold cross-validation. The mean accuracy across all subjects resulted in 93.07% and 96.14% using LDA and L-SVM, respectively. In summary, this research represents a further step forward in the field of FMG-based HMIs. The reduction of FMG sensors and optimization of data processing techniques, including segmentation and feature extraction, provide a more effective and easier-to-use HMI that can be employed in a variety of applications ranging from computer applications, games, prosthesis and exoskeleton control, etc.