Measurements comparison of finger joint angles in hand postures between an sEMG armband and a sensory glove

This study compares the simultaneous measurements of finger joint angles obtained with a myoelectric armband (Myo), composed of eight surface electromyography (sEMG) sensors mounted on an elastic support, and a data glove, equipped with ten flex sensor on metacarpal and proximal finger joints. The flexion angles of all finger joints in four hand postures, that is open hand, closed hand and grasping two 3D printed molds of different size, were measured with a manual goniometer, and used to create, for each finger joint, a linear model from the measurement of the corresponding flex sensor in an electronic glove, as well as a regression model from the simultaneous measurements of 8 sEMG sensors of the Myo armband. The regression models were extracted testing different algorithms from the Matlab Regression Learner Toolbox. The performance of the models of the two wearable devices were evaluated and compared, applying a standard test, taken from literature on sensory gloves to evaluate the repeatability, reproducibility and reliability of finger joint measurements. These results were also compared with those reported by published works that followed the same standard test, using data gloves based on different sensing technologies. This work aims to demonstrate that the sEMG armbands can be applied to register the static postures of each finger joint with almost the same accuracy of sensory gloves.