カンファレンス (国際) PredicTaps: Latency Reduction Technique for Single-taps Based on Recognition for Single-tap or Double-tap
Conference on Human Factors in Computing Systems (CHI 2020)
In general, a system with touch input waits for a certain period of time (typically 350 -- 500 ms) for a subsequent tap to determine whether the initial tap was a single tap or the first tap of a double tap. This results in latency of hundreds of milliseconds for a single-tap event. To reduce the latency, we propose a novel machine-learning-based tap recognition method called "PredicTaps". In the PredicTaps method, by using touch-event data gathered from the capacitive touch surface, the system immediately predicts whether a detected tap is a single tap or the first tap of a double tap. Then, in accordance with the prediction, the system determines whether to execute a single-tap event immediately or wait for a subsequent second tap. This paper reports the feasibility study of PredicTaps.