Joshua Tashman

Joshua Tashman


Student Year: 
G1
Graduate Program: 
Biomedical Engineering (CMU)
Advisor: 
Keith Cook, PhD
Thesis: 
Adam Feinberg, PhD
Undergraduate Institution: 
Cornell University
Research: 

We have recently implemented on a custom lower leg prosthesis a robust local feedback control system which produces natural human walking dynamics and leg kinematics while tolerating ground disturbances and slopes without parameter adjustments. This system takes as inputs two control variables, the landing target leg angle αtgt, and the minimum leg length during swing lclr. We seek to implement real-time EMG control of either αtgt or lclr on our prosthesis. To do so, we will collect several model data sets including 3D motion tracking, EMG measurements, and ground reaction forces on a split belt laboratory treadmill. We have the ability to control a walking subject’s αtgt and can extract lclr using 3D motion tracking data. In these data sets we expect to find a meaningful relationship between upper leg muscle EMG patterns and either one or both of the control variables. If such a relationship exists, we will produce a predictive model which will use real time EMG measurements to actively determine the value of one of the control variables. A new control system will then be developed to use this predictive model to modulate the prosthesis’ control. After this predictive model and control system have been transferred to and tuned on the prosthesis we will return to the treadmill to evaluate its performance. We expect that EMG modulated control of our prosthesis will reproduce typical human reactions to stumbling and in turn reduce the likelihood of its users falling.