A Supervisory Fuzzy-PID Controller for a MIMO Biped Robot Balance in Frontal Plane

Abstract

In this paper we propose to control a bipedal robot in an unstable position by means of a PID controller that gains are turned by a fuzzy logic system. For that, a model of planar 3 linked segment consisting of limb, trunk and extended arms with fixed base is used. Fuzzy if-then rules are constructed based on human expert knowledge and biomechanics studies for tuning of PID’s gain. For construction of tuning rules, we have developed an optical measuring system to record experimental data of balance keeping of a human in an unstable position. The control model is based on three sets of different global variables: (1) limb orientation and its derivative, (2) trunk/upper attitude and its derivative, and (3) orientation of extended arms and its derivative. In this study, we consider only side to side robot’s balancing and we also demonstrate, by simulation results, that the proposed control model is stable in presence of external disturbance. Finally we have compared the simulation results with the experiments to show the similarity of the proposed system with the human balance keeping task.