Application of different Control Strategies to the Forward Dynamic Simulation of Human Gait

Abstract

In this thesis, two different control strategies are applied to the forward dynamic simulation of multibody systems in order to track a given reference motion. For this purpose, two different computational models are presented: a four-bar linkage model with one degree of freedom; and a two-dimensional human body model that consists of 12 segments with 14 degrees of freedom. The forward dynamic analysis of the two models is implemented using the matrix-R formulation and carried out by means of a variablestep integration solver. Furthermore, an analysis and comparison of different numerical integration methods are carried out. The joint forces and torques, which are applied to the multibody systems in order to drive their motion, are provided through an inverse dynamic analysis. In order to stabilize the simulation and to enable the tracking of a reference motion, two control methods are introduced: a proportional derivative control and a computed torque control using feedback linearization. The design of both control approaches is developed and applied to the forward dynamic simulation of both models. The system performance is evaluated by comparing the results with the reference motion. The reference human motion of a healthy subject was captured previously in a biomechanics laboratory. Moreover, the robustness of the computed torque control approach is analysed. In addition, environmental and social impacts of this thesis are outlined and an economical consideration is included.