Faculty Advisor or Committee Member

Denis Rancourt, Committee Member

Faculty Advisor or Committee Member

Todd R. Farrell, Committee Member

Faculty Advisor or Committee Member

Donald Richard Brown III, Committee Member

Faculty Advisor or Committee Member

Edward A. Clancy, Advisor




"This compendium-format dissertation (i.e., comprised mostly of published and in-process articles) primarily reports on system identification methods that relate the surface electromyogram (EMG)—the electrical activity of skeletal muscles—to mechanical kinetics. The methods focus on activities of the elbow and hand-wrist. The relationship between the surface EMG and joint impedance was initially studied. My work provided a complete second-order EMG-based impedance characterization of stiffness, viscosity and inertia over a complete range of nominal torques, from a single perturbation trial with slowly varied torque. A single perturbation trial provides a more convenient method for impedance evaluation. The RMS errors of the EMG-based method were 20.01% for stiffness and 7.05% for viscosity, compared with the traditional mechanical measurement. Three projects studied the relationship between EMG and force/torque, a topic that has been studied for a number of years. Optimal models use whitened EMG amplitude, combining multiple EMG channels and a polynomial equation to describe this relationship. First, we used three techniques to improve current models at the elbow joint. Three more features were extracted from the EMG (waveform length, slope sign change rate and zero crossing rate), in addition to EMG amplitude. Each EMG channel was used separately, compared to previous studies which combined multiple channels from biceps and, separately, from triceps muscles. Finally, an exponential power law model was used. Each of these improvement techniques showed better performance (P<0.05 and ~0.7 percent maximum voluntary contraction (%MVC) error reduction from a nominal error of 5.5%MVC) than the current “optimal” model. However, the combination of pairs of these techniques did not further improve results. Second, traditional prostheses only control 1 degree of freedom (DoF) at a time. My work provided evidence for the feasibility of controlling 2-DoF wrist movements simultaneously, with a minimum number of electrodes. Results suggested that as few as four conventional electrodes, optimally located about the forearm, could provide 2-DoF simultaneous, independent and proportional control with error ranging from 9.0–10.4 %MVC, which is similar to the 1-DoF approach (error from 8.8–9.8 %MVC) currently used for commercial prosthesis control. The third project was similar to the second, except that this project studied controlling a 1-DoF wrist with one hand DoF simultaneously. It also demonstrated good performance with the error ranging from 7.8-8.7 %MVC, compared with 1-DoF control. Additionally, I participated in two team projects—EMG decomposition and static wrist EMG to torque—which are described herein. "


Worcester Polytechnic Institute

Degree Name



Electrical & Computer Engineering

Project Type


Date Accepted





biomedical signal analysis, joint impedance, EMG to torque, Electromyogram, EMG, motor units, decomposition