SVR CONTROLLER FOR A BIPED ROBOT IN THE SAGITTAL PLANE WITH HUMAN-BASED ZMP TRAJECTORY REFERENCE AND GAIT

Abstract

This paper introduces two new important issues to be considered in the design of the zero moment point (ZMP) trajectory of a biped robot. It was verified experimentally that in the human gait the ZMP trajectory moves along the foot in a way that it is shifted forward relative to its center. To take this into account a shift parameter is then proposed. It was also verified experimentally that in the human gait the ZMP trajectory amplitude depends on the swing time, reducing to zero for a static gait. It is then proposed a parameter to take into account this variation with the swing time of the gait. Six experiments were carried out for three different X ZMP trajectory references. In order to evaluate and compare the performance of the biped robot using the three X ZMP trajectory references two performance indexes are proposed. For the real-time balance control of this 8 link biped robot it was used an intelligent computing control technique, the Support Vector Regression (SVR). The control method uses the ZMP error and its variation as inputs and the output is the correction of the robot's ankle and torso angles, necessary for the sagittal balance of the biped robot.