Real-time Electromyogram-based Gesture Recognition using Machine Learning for Controlling Powered Wheelchair

This thesis enhances powered wheelchair control for motor-impaired individuals using sEMG signals and AI-based gesture recognition. Five gestures control unidirectional steering, achieving 95.50% accuracy in real-time navigation. For finger impairments, user-independent classification methods like MC-SVM and DT ensure high recognition accuracy. The study also introduces a human-machine interface utilizing an artificial neural network, achieving 99.06% accuracy and enabling collision-free navigation. This technology has the potential to empower individuals with disabilities, offering improved mobility and independence.