Skin-Conformal Myography for Real-Time Hand Tracking Using a Laser-Induced Graphene Strain Sensor Array

Despite significant progress in developing wearable systems for hand tracking, most devices are still bulky, restrictive to the user or to the placement of the exoskeleton systems, and sensitive to skin preparation and impedance. In this work, we develop a wristband that integrates an array of 10 skin-conformal strain sensors based on laser-induced graphene, which is optimized for continuous measurement of skin strain. The device is characterized to identify several hand gestures and tasks while simultaneously using an optical camera-based hand-tracking system to estimate the joint locations for ground truth generation. Machine learning models are developed to predict gestures as well as specific hand joint angles with high accuracy of >90% and >95%, respectively. The findings show that the sensors placed closer to actuation-specific anatomical features contribute more toward the high accuracy. The sensor array is also integrated with a wearable readout system that wirelessly transmits the data in real time in order to control a robotic arm as a proof of concept for human–robot interaction applications. The developed skin-conformal device is expected to find wide applications in rehabilitation, sports sciences, and human–computer interaction, paving the way for low-profile prosthetic and orthotic control systems.