Algorithmic Correction of Systematic Error in Eye Point-of-Regard (POR) Data Analysis

This project will develop and evaluate algorithms for correcting the systematic point-of-regard (POR) error found in the eye position data produced by modern eye-tracking hardware. The core idea of this project is the registration of graphical realities to the POR data. This registration enables the use of application-specific contextual information to correct systematic eye sensor error in static and dynamic displays. Phase I of the project will systematically develop and evaluate algorithms for POR error correction. The algorithms will detect operator behavior patterns by comparing expected application patterns to observed eye-movement data. The detected patterns will establish confident fixation locations (CFLs). The CFLs will provide the basis for the removal of systematic POR error. Phase II will further validate and implement the algorithms, categorize eye-tracking applications according to correction potential, and develop tools to support the implementation of POR correction in real-world applications.