Controlling Pivoting Gait Using Graph Model Predictive Control

Pivoting gait, in which a robot iteratively tilts an object, rotates it around a vertex, and then places it down on the floor, is efficient for manipulating a large and heavy object with a relatively small manipulating force. However, pivoting gait can easily fail, even with a small external disturbance, due to its instability. To address this problem, we propose a controller to robustly control the object's motion during pivoting gait by introducing two gait modes, i.e., double-support mode, which can manipulate a relatively light object with higher speed, and quadruple-support mode, which can manipulate a relatively heavy object with slower speed. To control the pivoting gait, a graph model predictive control is applied by considering these two gait modes. The experiments show that by adaptively switching the gait mode according to the applied external disturbance, a robot can stably perform the pivoting gait even when an external disturbance is applied to the object. The experimental results lead us to automate the manipulation of a large and heavy object.