Parameter adaptation using ground contact force for a humanoid walking control

In this paper, the ground contact force during a walking cycle is used as the feedback for walking parameter adaptation. The force is measured from piezo-resistive force sensors attached at the corner of the robot's footplate. The neural networks process the force measurement input and calculate the current observed leaning angle of the robot. The body leaning angle parameter is adjusted according to the observed leaning angle. With a good leaning angle parameter, the robot can walk with well balanced force on both feet and shows greater stability. The proposed process was tested with various initial conditions. The walking parameter controller can automatically adjust the leaning angle of the robot to the correct value which resulted in good walking performance.