Gaze Input for Ordinary Interfaces: Combining Automatic and Manual Error Correction Techniques to Improve Pointing Precision

Although eye tracking technology has greatly advanced in recent years, gaze-based interaction is still not as comfortable and effective as traditional mouse or touchpad input. In this paper we present the solutions we have developed to enhance eye pointing in ordinary interfaces, often characterized by small graphical elements. The described approach combines both automatic and manual error correction techniques (namely eye data filtering, virtual magnetization, magnifying glass and cursor manual shift) to allow the use of gaze as an alternative or supplementary communication channel. A modified keyboard with two additional keys is also proposed, with the purpose to speed up gaze interaction. Experiments have shown that the approach can provide better performance than touchpad pointing and compares fairly with the mouse. The obtained results, while preliminary, can be the starting point for deeper and more focused investigations, driving further research on the topic.