Abstract: By using Model Predictive Impedance Control (MPIC), two essential
questions about human walking control have been considered in this paper: Which
parameters are controlled by the control system and what strategies are
applied? As a general method, MPIC was proposed as a possible model for
controlling human movements in the previous paper of one of the authors [1]
and it was also applied to periodic movements such as walking [2]. According
to the acceptable results of primary experiences, a more realistic walking
model is presented in this paper to be controlled by MPIC strategy. A
3-dimensional 5-link simulated walking robot with 8 degrees of freedom in
sagittal and lateral planes is employed to be analyzed while being controlled
on an irregular condition, like ascending and descending stairs. The objective
of the controller is to optimize the energy consumption, vertical orientation
of the body, and forward speed of its center of mass. Visual feedback effect is
applied when the robot targets specific footprints over the walking terrain.
Each optimization sequence has been applied during the prediction horizon with
one human step duration. The results have been animated for comparison with
natural walking of a human in similar conditions.
Keywords: Model predictive impedance control, biped walking model, rough terrains
