Influence of the controller design on the accuracy of a forward dynamic simulation of human gait

The analysis of a captured motion can be addressed by means of forward or inverse dynamics approaches. For this purpose, a 12 segment 2D model with 14 degrees of freedom is developed and both methods are implemented using multibody dynamics techniques. The inverse dynamic analysis uses the experimentally captured motion to calculate the joint torques produced by the musculoskeletal system during the movement. This information is then used as input data for a forward dynamic analysis without any control design. This approach is able to reach the desired pattern within half cycle. In order to achieve the simulation of the complete gait cycle two different control strategies are implemented to stabilize all degrees of freedom: a proportional derivative (PD) control and a computed torque control (CTC). The selection of the control parameters is presented in this work: a kinematic perturbation is used for tuning PD gains, and pole placement techniques are used in order to determine the CTC parameters. A performance evaluation of the two controllers is done in order to quantify the accuracy of the simulated motion and the control torques needed when using one or the other control approach to track a known human walking pattern.