Optimization of ankle stiffness using genetic algorithm in gait modeling

In the human gait modeling, it is common to employ 2D models that consist of a chain of rigid links joined together by frictionless hinge joints. Although Newton’s method is usually used to obtain equations of motion in the previous studies, in this research, the constrained Lagrange’s method was employed for this purpose. This method has some advantages over the previous one, such as the solution process is independent of the coordinate system and there is no necessity to know the ground reaction force beforehand. In this work, optimization was also performed by genetic algorithm so that the moment of each joint was estimated by tracking the kinematic data. Moreover, by solving the inverse dynamics and by applying Lagrange multipliers, the distribution of ground reaction force under both feet in the double support mode was calculated and compared with the experimental data to verify the effectiveness of the proposed method. Finally, as one of the applications of dynamic modeling of the human gait, the optimal value of passive stiffness in the ankle joint was obtained to provide a better design of the orthoses used for patients with motor impairment. The results show compatibility between the simulations and experiments for normalized joint moments as well as reaction forces. The optimal joint stiffness is also in the range reported by available experimental data. In conclusion, the methodology can be used for modelling human movements and can be considered as an optimal approach in designing assistive devices especially passive exsoskeletons.