A novel, co-located EMG-FMG-sensing wearable armband for hand gesture recognition

Abstract

Gestures play an important role in human-computer interaction, providing a potentially intuitive way to bridge the gap between human intention and the control of smart devices. Electromyography (EMG) and force myography (FMG) are two commonly-adopted wearable sensing modalities for gesture recognition. Previous research approaches utilize only a single modality (EMG or FMG) at any given muscle location, thus limiting the amount of potentially-useful biological hand gesture information. We thus propose a novel co-located approach (EMG and FMG) for capturing both sensing modalities, simultaneously, at the same location. We developed a novel hand gesture recognition armband consisting of 8 co-located EMG-FMG sensing units (size and weight of each EMG-FMG sensing unit was 11 × 13 × 6 mm and 0.90 g, respectively). Five subjects performed a hand gesture recognition experiment for American Sign Language digits 0–9 while wearing the co-located EMG-FMG armband on the forearm. Hand gesture classification accuracy was 81.5 % for EMG only, 80.6 % FMG only, and 91.6 % for co-located EMG-FMG. These results suggest that co-located EMG-FMG may lead to higher hand gesture classification accuracy than sensing approaches using either EMG or FMG in isolation. To the best of our knowledge, this is the first prototype that measures EMG and FMG simultaneously at the same muscle location for hand gesture recognition. Implications of this work could positively impact a variety of muscle activity monitoring research applications including biomechanics modeling, prosthesis control, and gesture recognition.