Date of Award

Spring 1-1-2021

Degree Type


Degree Name

Doctor of Philosophy In Mechanical Engineering Degree


Mechanical Engineering

First Advisor

Richter, Hanz

Second Advisor

Antonie van den Bogert, Ph.D

Third Advisor

Eric Schearer, Ph.D


This dissertation focuses on the study and optimization of human training and its physiological effects through the use of advanced exercise machines (AEMs). These machines provide an invaluable contribution to advanced training by combining exercise physiology with technology. Unlike conventional exercise machines (CEMs), AEMs provide controllable trajectories and impedances by using electric motors and control systems. Therefore, they can produce various patterns even in the absence of gravity. Moreover, the ability of the AEMs to target multiple physiological systems makes them the best available option to improve human performance and rehabilitation. During the early stage of the research, the physiological effects produced under training by the manual regulation of the trajectory and impedance parameters of the AEMs were studied. Human dynamics appear as not only complex but also unique and time-varying due to the particular features of each person such as its musculoskeletal distribution, level of fatigue,fitness condition, hydration, etc. However, the possibility of the optimization of the AEM training parameters by using physiological effects was likely, thus the optimization objective started to be formulated. Some previous research suggests that a model-based optimization of advanced training is complicated for real-time environments as a consequence of the high level of v complexity, computational cost, and especially the many unidentifiable parameters. Moreover, a model-based method differs from person to person and it would require periodic updates based on physical and psychological variations in the user. Consequently, we aimed to develop a model-free optimization framework based on the use of Extremum Seeking Control (ESC). ESC is a non-model based controller for real-time optimization which its main advantage over similar controllers is its ability to deal with unknown plants. This framework uses a physiological effect of training as bio-feedback. Three different frameworks were performed for single-variable and multi-variable optimization of trajectory and impedance parameters. Based on the framework, the objective is achieved by seeking the optimal trajectory and/or impedance parameters associated with the orientation of the ellipsoidal path to be tracked by the user and the stiffness property of the resistance by using weighted measures of muscle activations.