Character Input Interface Design for Effective Use of Motion Sensors on Smartwatches

Abstract

In recent years, the opportunity to operate smartwatches has increased owing to the increase in the number of people using smartwatches. The main input methods for smartwatches are touch and voice; however, there is/are these input methods present a few problems. In this study, we focus on gesture input, which involves detection via hand and arm movements. It can be discriminated by the accelerometer and gyroscope in a smartwatch. Previously, we attempted to discriminate finger movements; however, each finger movement was difficult to distinguish. This implies that discriminating gestures using motion sensors is challenging. Therefore, we herein propose side-tap, dial-rotation, and band-rotation operations, which can be effectively performed in a motion sensor. The purposes of these operations are to tap the side of the device, twist the dial, and twist the wrist, respectively. The characteristic waveforms of these operations can be obtained using the accelerometer and gyroscope in the smartwatch. In addition, the proposed operation offers a sufficient number of selections for the character input. The waveforms of the proposed operations are analyzed, and a discrimination algorithm is developed and evaluated. First, each motion is performed 10 times consecutively; subsequently, each motion is randomly performed 10 times. The results show that the average discrimination rate is 97.5%. Therefore, the proposed motion is useful as a new input method for smartwatches. In the future, to use the proposed motion as a character input, an appropriate keyboard design is necessitated. Therefore, we develop a keyboard design that enables users to input characters intuitively using the proposed motion.