Wheelchair Control Based on EOG Signals of Eye Blinks and Eye Glances Based on The Decision Tree Method
Muhammad Ilhamdi Rusydi, Muhammad Abrar A Boestari, Riko Nofendra, Syafii Syafii, Agung Wahyu Setiawan, Minoru Sasaki · 2024
This research investigates the development of an EOG (electrooculography)-based wheelchair controlled by eye movements, including blinks and glances. The system leverages peak signal features for conscious blinking (on/off control), unconscious blinking (standing still), and upward glances (moving forward). Additionally, it utilizes signal polarity features for left/right glances (turning) and downward glances (moving backward). Three decision tree classification models were generated, and the model with post-pruning was chosen due to its simpler structure, superior test data accuracy, and resistance to overfitting. Testing with 15 participants yielded promising results. Wheelchair movement accuracy reached 98.89%, and track-based testing demonstrated smooth movements with an accuracy of 96%. These findings suggest the potential of EOG-based control for wheelchairs, offering increased independence and improved quality of life for users with limited mobility.