Stiffness Control of an Active Transtibial Prosthesis

Abstract

Active, transtibial prostheses typically use finite state control algorithms that struggle with cadence and gait variability of the amputee. Recent work in artificial neural networks (ANN) have shown the possibility to predict the users intent based on EMG activity and the current position of the ankle, which can be used as an input signal into an improved controller. This paper examines how to implement an ANN signal into a zero order impedance controller, i.e., a stiffness controller, on a specific active transtibial prosthesis. The prosthesis incorporates a linear spiral spring in parallel with a four-bar mechanism. In order to implement stiffness control, the spring was moved to being in series with the four-bar mechanism to establish a relationship between the torque of the spring and the position of the motor. To ensure stiffness control is feasible, a MATLAB Simulink model of the system was created to test the robustness of the controller and the effect of moving the spring from parallel to series. The robustness of the controller was verified as the ankle position and torque requirements are met in the simulation. The Simulink model accurately models the new system and can be used in the future to optimize the motor or the four-bar mechanism for this new type of control.