The adoption of high-density electrode systems for human–machine interfaces in real-life applications has been impeded by practical and technical challenges, including noise interference, motion artefacts and the lack of compact electrode interfaces. To overcome some of these challenges, we introduce a wearable and stretchable electromyography (EMG) array, and present its design, fabrication methodology, characterisation, and comprehensive evaluation. Our proposed solution comprises dry-electrodes on flexible printed circuit board (PCB) substrates, eliminating the need for time-consuming skin preparation. The proposed fabrication method allows the manufacturing of stretchable sleeves, with consistent and standardised coverage across subjects. We thoroughly tested our developed prototype, evaluating its potential for application in both research and real-world environments. The results of our study showed that the developed stretchable array matches or outperforms traditional EMG grids and holds promise in furthering the real-world translation of high-density EMG for human–machine interfaces.
Surface electromyography (sEMG) has long been employed in basic neuroscience research and medical applications, such as in prosthetic control and neuromuscular diagnosis. More recently, major efforts have been directed toward extending this technology to everyday-life interfaces, exemplified by the development of commercial neural-interface wristbands. However, the control bandwidth of current systems remains fundamentally constrained. This limitation arises from the difficulty of learning representations that capture fingerspecific degrees of freedom in the highly overlapping sEMG signal space, as well as from the scarcity of large annotated datasets for training deep models. Here, we propose a multiencoder neural architecture that learns a disentangled latent representation of finger activity from sEMG during isometric contraction gestures. The model is trained using only a limited subset of single- and multi-finger gestures, yet it generalizes to previously unseen combinations in a zero-shot learning setting. Evaluated on recordings from 8 healthy individuals, the method achieved an average accuracy of 88.74% on all gestures (both seen during training, as well as unseen gestures), significantly outperforming both standard and disentanglement-based deep learning baselines by more than 30%. By enabling robust decoding of complex finger movements from limited training data, this architecture paves the way for widespread adoption of neural interfaces in everyday life.
This paper presents different methods used to allow a number of radar sensors mounted in a room to derive their relative spatial position as well as the floor plan of the room itself by relying on measurements of the distance of a person (walking around the perimeter of the room) from each sensor, at various time instants.