Smooth Pursuit Target Speeds and Trajectories

In this paper we present an investigation of how the speed and trajectory of smooth pursuits targets impact on detection rates in gaze interfaces. Previous work optimized these values for the specific application for which smooth pursuit eye movements were employed. However, this may not always be possible. For example UI designers may want to minimize distraction caused by the stimulus, integrate it with a certain UI element (e.g., a button), or limit it to a certain area of the screen. In these cases an in-depth understanding of the interplay between speed, trajectory, and accuracy is required. To achieve this, we conducted a user study with 15 participants who had to follow targets with different speeds and on different trajectories using their gaze. We evaluated the data with respect to detectability. As a result, we obtained reasonable ranges for target speeds and demonstrate the effects of trajectory shapes. We show that slow moving targets are hard to detect by correlation and that introducing a delay improves the detection rate for fast moving targets. Our research is complemented by design rules which enable designers to implement better pursuit detectors and pursuit-based user interfaces.