Validation of methods for determining ankle stiffness during walking using the Perturberator robot

Abstract

Recently developed powered ankle prostheses are capable of providing users with biologically inspired control during walking. However, currently, the appropriate dynamic mechanical properties, or impedance, of the human ankle during walking is unknown. Before trustworthy estimates of the ankle's impedance can be obtained using the Perturberator robot, it must be thoroughly validated. In this study, the sensitivity of standing ankle impedance estimates to foot placement was investigated. Additionally, linear filters that mapped acceleration of the Perturberator motor angle to the forces caused by the robot's intrinsic impedance were determined. Lastly, impedance estimates of a prosthetic foot were obtained at four perturbation timing points during the stance phase of walking and compared to values obtained from an independent measure of prosthetic ankle stiffness. During standing, foot placement had a significant effect on ankle impedance measurements (p < 0.001). The linear filters accounted for, on average, 98% of the variance in the forces caused by a perturbation. Lastly, when the impedance of the prosthetic foot was determined during walking, there was 3% error when compared to the stiffness measured by the independent measure at the appropriate timing in stance phase. This work was a preliminary, but important step toward our goal of determining the impedance of the human ankle during walking.