Subject-Independent EOG-Based Classification of Horizontal Eye Movements for Real-Time Robot Control

Background/Aim: Neurodegenerative diseases and severe neurological trauma can substantially impair conventional communication and control channels, while ocular motor functions may remain preserved until advanced stages. This study proposes a subject-independent electrooculography (EOG)-based framework for classifying horizontal gaze directions and controlling an omnidirectional mobile robot without user-specific calibration.Methods: EOG signals acquired from 12 healthy participants were filtered using a fourth-order Butterworth band-pass filter between 0.1 and 15 Hz and segmented into non-overlapping 5-second windows. Twelve time- and frequency-domain features were extracted from each window. Five classifiers—k-Nearest Neighbors (KNN), Random Forest (RF), Support Vector Machine (SVM), Decision Tree (DT), and a multilayer perceptron-based artificial neural network (ANN)—were evaluated using Leave-One-Subject-Out cross-validation.Results: SVM achieved the highest participant-wise mean accuracy of 79.61%, followed by KNN (74.47%), RF (72.19%), DT (68.69%), and ANN (66.11%). Pooled out-of-fold analysis further showed that SVM obtained a macro specificity of 0.911 and a macro one-vs-rest ROC-AUC of 0.907. The classified gaze commands were transmitted wirelessly to the mobile robot for real-time motion control.Conclusion: The results demonstrate that the proposed framework provides a practical and computationally efficient solution for subject-independent assistive robotic control.