A Wearable Human–Machine-Interface (HMI) System Based on Colocated EMG-pFMG Sensing for Hand Gesture Recognition

There have recently been some significant research activities on sensors or electrodes for human–machine-interface (HMI) systems, mainly focusing on single-modality sensing systems. Colocated multimodal sensing systems are employed to improve the performance of such a sensing system and eliminate the limitations of single-sensor-based HMI for gesture recognition, for example, to control prosthetic devices. Therefore, we propose a new low footprint armband-based HMI system consisting of colocated electromyography (EMG) and pressure-based force myography (pFMG) sensors. What is different from the literature is to employ pressure-sensitive air chambers as the pFMG system on which the EMG electrodes are strategically placed to operate simultaneously to measure the electrical and mechanical signals from the same site. The soft pneumatic sensing chamber with a reasonable size (45 × 20 × 12 mm) has been designed and 3-D-printed using flexible filaments for this purpose. The system consists of eight pairs (not necessarily the only option) of EMG-pFMG sensors, providing signals with a sampling frequency of 1000 Hz. The system was tested with 14 subjects with an average accuracy of 94.6% for the recognition of seven commonly used gestures, which was proved to be higher than EMG-only (87.95%) and pFMG-only (80.11%), using machine learning techniques. When a minimum number of sensors is used, the advantage of multiple sensing modalities is even more obvious. The results presented demonstrate the effectiveness of the HMI system based on the colocated EMG-pFMG sensing techniques in enhancing the performance of a gesture recognition system.