Mimicking 2-DOF Human Eye Movement using an IR Camera-Based Wearable Eye-Gaze Tracking Controller

摘要

This study presents the real-time human-robot mimicry of the human eye two degrees-of-freedom (2-DOF) movements. The control system uses an IR camera-based wearable eye-gaze tracking controller attached to the human head to control the robotic eye. Building upon previous studies, this study addresses limitations of existing eye gaze tracking systems for robotic control in horizontal and vertical eye movements including attempts to compensate for head movements using a wearable controller. The researchers employ image processing through IR camera in eye-gaze tracking system in the wearable controller, applying the Pupil Center Corneal Reflection (PCCR) concept and polynomial regression that enables to replicates the eye movements made by humans into the robotic eye. After a statistical analysis done in the data gathered through Wilcoxon Test, the p-values obtained for horizontal and vertical movements are higher than the alpha level. This means that the observations indicate no statistically significant difference between the angles of the human eye and the robotic eye. The results demonstrate that the wearable eye-gaze tracking controller’s control system allows the robotic eye to mimic human eye movements, with a highly accurate horizontal movement of 99.07% and vertical movement of 98.91%, resulting in an overall accuracy of 98.99% in mimicking human eye movements.