High-Speed Gesture-Cognitive Exo-Glove via Electrostiction

In this study, we developed an artificial intelligence (AI) exo-glove that changes the knuckle stiffnesses quickly and accurately in harmony with user’s intention, and thus, assists or controls hand gestures. Electrostiction phenomenon based on Maxwell stresses in a high dielectric film (HDF) can produce high attractive force between two HDF/electrode/substrate layers. The high attractive force in turn can induce high frictional force between the two contacting layers, and thus, can change the linear and rotational stiffnesses between two linked layers. When a poly(vinylidene fluoride-trifluoroethylene-chlorotrifluoroethylene) [P(VDF-TrFE-CTFE)] thin film of thickness 10 μm is used as the high dielectric films, a frictional shear stress of 27.92 N/cm² is induced under applied voltage of 250 V. Based on electrostiction mechanism, an exo-glove composed of multiple multilayered links are developed in this study, which fit the outer shape of five fingers and freely move together with hand gesture.<br/>In order to perceive human’s intention and cooperate the exo-glove with human’s hand gesture, 8-channel fabric-type electromyography (EMG) sensors are used. EMG signals collected from a user in real-time through the EMG sensors are analyzed and classified into several gesture patterns with the support vector machine (SVM) algorithm. This classification (decision) is converted into voltage signals that control the multiple multilayered links in the exo-glove and therefore control the hand gesture by changing the stiffnesses at high speed in harmony with one’s intention. As a demonstration of the AI exo-glove developed in this study, two persons (one (A) wears the AI exo-glove, and the other (b) wears the EMG sensors) perform rock-paper-scissors games, and the one (A) wearing the AI exo-glove can mostly win the rock-paper-scissors games by aid of the machine learning algorithm and exo-glove. This demonstration shows how an exo-glove and also exo-suit operated by electrostiction mechanism can cooperate with a user to assist his or her gesture and motion by using a machine learning algorithm.