Fast Target Selection via Saccade-driven Methods

Abstract

Three fast, saccade driven, target selection methods are explored. The first method selects a target at the beginning of a saccade with the objective of providing target selection in almost constant amount of time regardless of the distance to the target. The second method selects a target at the end of a saccade. The third is a hybrid method combining the speed of the saccade-driven selection with the accuracy of the conventional Dwell-Time selection. Theoretical evaluation of the proposed methods conducted via characteristics of the Human Visual System and a mathematical model of the human eye indicates that the objective is tenable. Practical evaluation of the proposed methods is conducted with the Multi-Directional Fitts' Law task and with a real-time eye-gaze-guided video game designed to simulate gaming environments where selection speed of a target is of outmost importance. The results indicate that proposed methods show an increased throughput and task completion performance compared to the conventional Dwell-Time target selection method.