Observer-Based Postural Balance Control for Humanoid Robots

Maintaining postural balance is a fundamental requirement for humanoid robots faced with unknown terrain or external perturbations. In this paper we present an observer-based balancing control strategy that is robust to persistent perturbations on the ground slope which is based on the estimation of the disturbance without the need for an inertial measurement unit. We use the workspace control method in combination with an observer to perform specific maneuvers while maintaining postural balance. We employ a linear observer, a gain-scheduling observer and an extended Kalman filter and compare the results obtained. For the control task considered, while the linear observer falls short in this setting, both the gain-scheduling observer and the extended Kalman filter perform well, which is confirmed by experimental results on a planar biped robot.