Forward dynamics simulation and control of dorsiflexion and planterflexion of the human foot

Abstract

The thesis uses the forward-dynamics approach to calculate the forces of the tibialis anterior and soleus muscles by controlling the dorsiflexion and plantarflexion movements of the ankle joint. The muscles are modeled with the Hill-type muscle model and are controlled either using a Proportional-Integral-Derivative (PID) controller or Model Predictive Control (MPC). Another simulation includes the replication of the ankle dynamometer experiment and MPC is used to control the trajectory. The study demonstrates that the MPC provided superior accuracy in controlling the desired trajectory and is a good option when nonlinearities are involved. Moreover, the ability of MPC to have constraints enables it to predict the states of the system. However, MPC is significantly slower in terms of the execution time of the simulation as it requires higher computational cost. On the other hand, results also show that the computational cost can be lowered by reducing the horizon length of the MPC, reducing the maximum iterations and acceptance tolerance of the nonlinear solver, etc. However, there will be a trade-off between accuracy and the computational cost of the system. The best way to reduce the computational cost of MPC is to involve the least amount of states in the model. This is possible by removing the redundant states and using the minimal description instead, which will be considered in the future.