Machine learning-augmented wearable triboelectric human-machine interface in motion identification and virtual reality

Motion detection from flexible and self-powered human-machine interfaces (HMI) is a promising research field for the Internet of Things (IoT) and virtual reality. The traditional HMI normally generated a signal curve that used the point value to represent motion status. With the proposed dual-electrode setup scheme, this research reported machine learning-augmented, wearable, self-powered, and sustainable HMI sensor in human finger motion and its virtual activities. The triboelectric friction between the moving object and the specific electrode array was used to regulate the programmable output curve of the instantaneous parameters with a unique and stable electrical signal. It demonstrated that the movable object’s motion can be tracked and accurately reproduced by the output signal using a decoupling algorithm into different motion patterns. Furthermore, the machine learning algorithm can classify fast and slow finger motions with obviously visualization performance. It also demonstrated that multiple linear regression (MLR) and the principal component analysis (PCA)+K-means clustering (K-means) presented substantially in terms of clustering, visualization, and motion speed interference. This research not only established the viability of designing self-powered HMI sensors but also demonstrated a way for identifying machine learning-augmented motion patterns and potential virtual activities by self-powered, wearable triboelectric HMI sensors.