Virtual Keyboard Based on MEMS Sensor Network and Fusion of Accelerometer and Gyroscope

The paper reports a novel technique in the field of human-computer interaction based on MEMS Inertial Measurement Unit (IMU). It enables inputting the characters by using four wearable sensors without the actual physical contact with a keyboard. In this paper, a virtual keyboard scheme is presented and the keystrokes are recognized through fusion data of accelerometer and gyroscope. We developed the sensor network using four MEMS sensors (MPU6050)and a microcontroller (STM32F103). The attitude angle of each finger during the keystroke are calculated from the data fusion using the angle compliment filter algorithm. And then, the different keystrokes are recognized based on the attitude angle of fingers. Finally, the Support Vector Machine (SVM)algorithm was adopt for recognizing the keystrokes, and the finial recognition correct rate can achieve 95.07%. Index Terms-Wearable Sensor, Inertial Measurement Unit, keystroke recognition, Virtual Keyboard