Fig. 4. Demonstration of a wearable p-NHE for wireless, multi-class human-machine interfaces.
a Schematic illustration (top) showing target muscles on forearm to recognize multiple gestures and photos (bottom) capturing three p-NHE positioned on targeted muscles, including palmaris longus, brachioradialis, and flexor carpi ulnaris. Enlarged image shows one of those systems with a circuit and electrodes. b EMG mapping data showing RMS signals from multiple channels that cover the entire forearm; six different heat maps are obtained according to six gestures, including hand closed, thumb, index, middle, ring, and little finger motion (from left photo to right photo), for determining the ideal channel locations. c Representative EMG signals from four gestures, including open hand, closed hand (left graphs) and index finger flexion and wrist flexion (right graphs). d Demonstration of EMG-enabled human-machine interfaces with a wireless, wearable p-NHE to precisely control a flying drone (left), RC car (middle), and presentation software (right). In this test, a single-channel device is mounted on the muscle, palmaris longus (shown in a). e 3D plot of three-channel, EMG RMS signals for clear differentiation of seven different gestures as seven groups. f Summarized result of real-time, confusion matrix from ten trials, showing 98.6% accuracy across seven classes with three-channel EMG recording. g Demonstration of a three-channel EMG recording and corresponding control of a robotic hand, showing examples of six motions of a subject and following motions of the robotic hand.