SPEED: prédiction de cibles

Abstract

We present the SPEED method to predict endpoints, based on analysis of the kinetic characteristics of the pointing gesture. Our model splits the gesture into an acceleration phase and a deceleration phase to precisely detect target. The first phase allows us to identify a velocity peak that marks the beginning of the second phase. This phase is approached with a quadratic model to predict gesture endpoint. A pilot study shows that SPEED predicts a target more precisely than other existing methods, for 1D tasks without distractors.