AngularMomentum-Based Balance Control

Abstract

Maintaining balance under dynamic external disturbances and environmental conditions remains a key challenge of humanoid robots. As the centroidal dynamics of humanoid states that both linear and angular momenta must be regulated to completely control the balance, the momentum-based balance control approaches maintain the balance through controlling both the linear and angular momenta of a robot. In this setting, the joint motion of a humanoid robot is typically controlled to realize the desired momentum rate change while satisfying non-slip constraints for the support feet as well as some task objectives such as the desired posture. After reviewing related work, we establish the basic theories that compute centroidal momentum and its relation with the generalized coordinates of a humanoid robot. Then we introduce approaches to controlling momentum, such as the resolved momentum control and computed torque-based control. Subsequently, we present several momentum-based approaches to maintaining humanoid robots’ stationary balance in detail. Several ideas to set the desired angular momentum are presented as well. The chapter concludes with the discussion of the limitations and open questions for the momentum-based balance control.