Posture and balance control for humanoid robots in multi-contact scenarios based on Model Predictive Control

This work presents a new approach to whole-body control for balancing and posture stabilization of humanoid robots utilizing an optimization of contact forces in combination with Model Predictive Control. To perform tasks that require multiple contacts, like manipulation or crawling, the controller allows the robot to use a subset of its end effectors to apply a desired wrench to the environment. The remaining end effectors are used for balancing while taking into account the wrenches originating from the manipulation. Due to the prediction the controller is able to react to changes in the control inputs in advance. This approach is evaluated in simulation with the humanoid robot TORO.