EyeAssist: A communication aid through gaze tracking for patients with neuro-motor disabilities

Abstract

The paper presents EyeAssist, a low cost, user friendly communication aid developed specifically for individuals suffering from neuro-motor problems. Existing gaze based typing systems encounter problems of unintentional key presses, extensive calibration process and inbuilt noise of eye trackers. We tried to solve these problems in the EyeAssist system by employing accurate gaze tracking. Fixation information are derived from the acquired gaze data. The systematic and variable noises of the eye tracker are being removed by graph signal processing based filters, Kalman filter and linear transformation. A unique one-time calibration which is subject and session independent has made the system easy to use without compromising its efficacy. The system has provision for setting variable dwell time according to users' preferences. EyeAssist also provides some unique features like text to speech conversion of typed sentences, predictive word and sentence entry etc. Results show that the average typing rate is 11.8 words per minute with an average error rate of 6.6%. Moreover the error rate reduces drastically with some practice. The system is user friendly, easy to use which makes it a perfect choice for patients suffering with neuro-motor disabilities.