Analyzing eye tracking data using a Markovian framework to assess differences in scan patterns

Abstract

Data overload, especially in the visual channel, represents a major challenge with regards to display design in data-rich domains. One promising means of addressing data overload is with the use of eye tracking technology to better understand an operator’s transition methods between tasks in order to support operators in real-time. The goal of this study is to develop a Markovian framework analyzing eye movement across different panels while performing a simulated Unmanned Aerial Vehicle (UAV) control task, the chosen application of this study. Across ten participants, an increase in workload adversely affected performance, but did not change individual scan patterns, which were based on a Markovian framework. However, across participants, eye tracking data revealed five distinct scan patterns, each with varying levels of success in terms of response time and accuracy. The top four performers all adopted different scan patterns. The findings show that eye tracking can provide unique insights to explain performance differences between individuals. Overall, the findings provide the foundation for developing an algorithm that optimizes performance while accounting for individual differences.