A Practical Gesture-Based Interface for Real-Time Computer Control
via Smartphone Inertial Measurement Units

Abstract

This paper presents "IMU Gesture," a hands-free gesture interface for PC interaction utilizing smartphone IMU data—including orientation, angular velocity, acceleration, and barometric pressure—captured from a breast pocket. The system transmits sensor values via HTTP to a PC, where it sequentially identifies Basic, Roll, and Free gestures after preprocessing, employing template matching for free-form gesture recognition. To mitigate accidental inputs, we implemented a "Safety Gate" for managing high-risk operations, complemented by dry-run logging and a Web UI. A baseline recognition test and a user study (N=8) involving document editing tasks in Microsoft Word revealed that false positives and confirmation overhead remain challenges. However, the findings suggest that the interface is particularly viable for single-trigger tasks, such as presentation slide navigation, rather than complex document editing.