Click-based ultrasonic gesture recognition

Abstract

To maximize the potential of future human-computer interaction, advancements in the methods by which people interface with computers must be made. This study aims to establish the foundation for a new, contactless method of input powered by ultrasonic echolocation. The project consists of three main sub-studies; one short COMSOL study, and two rounds of data collection at different levels of detail. We built and analyzed COMSOL models in order to gain an initial sense of the feasibility of ultrasonic gesture recognition. The detailed data collection trials demonstrated a clear correlation between ultrasonic echoes and motion. During the rough data collection trials, we laid the groundwork for future endeavors which aim to use machine learning to build ultrasonic gesture recognition systems. Analysis of the data gained from the rough data collection trials made the preliminary conclusions of the detailed data collection trials more concrete..