Visually Perceived Relevance of Objects reveals Learning Improvements and Task Difficulty

Abstract

Recently, there has been increased interest in eye tracking and eye-based human-computer interaction research, as gaze reveals attention and intention information, making it highly interesting as an input modality for cognitive assistant systems. In this paper, we present how to apply gaze-awareness in an industrial environment to create such a cognitive assistant system. A prototypical system was built upon a mobile eye tracker, that analyses the user's field of view, to recognize possible objects of interest, while simultaneously checking for gaze fixations. A twenty-one participants user study was carried out, which builds upon established hypothesis of a correlation between eye fixations and the relevance of objects and additionally shows that this relationship changes over multiple tasks and runs, revealing learning improvements and task difficulty. Both features are vital for a reduction of amount of the support an assistance system provides towards its users, resulting in an overall better user experience.