Influence of Uncertainty in Metabolic Dynamic Time Constant on Instantaneous Cost Mapping Techniques

Introduction Newly developed ‘body-in-the-loop’ optimization algorithms have the potential to dramatically improve the performance of robotic assistive devices, such as prostheses and exoskeletons. These algorithms minimize a physiological cost function (e.g., energy expenditure) over a range of parameter values (e.g., controller timing) to determine the optimal parameter setting [1]. Successful implementation of these algorithms depends on the underlying instantaneous metabolic cost (i.e., how metabolic expenditure varies as a function of parameter setting). To this end, the relationship between instantaneous energetic cost, x, and experimentally collected breath measurements, y, can be modeled as a first-order linear system with a single subjectspecific time constant, τ , according to [1, 3]: