The development of current prostheses and orthoses typically follows a trial and error approach where the devices are designed based on experience, tried on human subjects and then redesigned iteratively. This design approach is costly, risky and time consuming. A predictive human gait model is desired such that prostheses can be virtually tested so that their performance can be predicted qualitatively, the cost can be reduced, and the risks can be minimized. The development of such a model is explained in this paper. The developed model includes two parts: a plant model which represents the forward dynamics of human gait and a controller which represents the central nervous system (CNS). The development of the plant model is explained in a different paper. This paper focuses on the control algorithm development and able-bodied gait simulation. The controller proposed in this paper utilizes model predictive control (MPC). MPC uses an internal model to predict the output in advance, compare the predicted output to the reference, and optimize control input so that the error between them is minimal. The developed predictive human gait model was validated by simulating able-bodied human gait. The simulation results showed that the controller is able to simulate the kinematic output close to experimental data.

This content is only available via PDF.
You do not currently have access to this content.