Quadratic program based control of fully-actuated transfemoral prosthesis for flat-ground and up-slope locomotion

Abstract

This paper utilizes a novel optimal control strategy that combines control Lyapunov function (CLF) based model independent quadratic programs with impedance control to achieve flat-ground and up-slope walking on a fully-actuated above-knee prosthesis. CLF based quadratic programs have the ability to optimally track desired trajectories; when combined with impedance control-implemented as a feed-forward term-the end result is a prosthesis controller that utilizes only local information while being robust to disturbances. This control methodology is applied to a bipedal robot with anthropomorphic parameters "wearing" a fully-actuated transfemoral prosthesis. Traditional human-inspired control methods are applied to the human component of the model-simulating nominal human walking-while the novel control method is applied to the transfemoral prosthesis. Through simulation, walking on flat-ground and up-slope is demonstrated, with the resulting gait achieved using the novel prosthesis control yielding walking that is nearly identical to the “healthy” human model. Robustness tests indicate that the prosthesis controller can endure large uncertainties and unknown disturbances.