Justin Phan

Title: Developing the Next-Generation Neural-Machine Interfaces for Neurorehabilitation Applications by Utilizing Sensor Arrays and Spatial Features

People with limb impairment or amputation face a severe reduction in quality of life, due to a hampered ability to perform daily activities. The current applications of machine learning to recognize movement intents in surface electromyography (sEMG) signals have shown promise in improving quality of life, by allowing intuitive control of assistive robotics. However, high rates of rejection have been reported due to lack of portability and functionality in existing systems. There is a need to improve the prediction accuracy of the pattern recognition (PR) algorithms that control myoelectric rehabilitation applications, while minimizing computational complexity and reducing hardware requirements for developed systems. The goal of this research project was to improve neuromuscular communication between humans and myoelectric machine interfaces, by utilizing a set of newly developed computationally efficient spatial features in PR algorithms. This strategy was implemented in the myoelectric control of a robotic arm to evaluate robustness in practical application.