Swing-up control, that is the transfer of a pendulum from a pendant state to the inverted one, is a good laboratory experiment of optimal control theory for non-linear control systems. The optimal control can be determined by the maximum principle and obtained as a function of time. Since the control is, however, determined in a feedforward fashion, the control is not robust to disturbances and uncertainties of the system, and the transfer of the state of the pendulum is not assured. In the paper, a robust swing-up control using a subspace projected from the whole state space is proposed. Based on the projected state space or pseudo-state, the control input is determined depending on the partitioning of the state as a bang-bang type control. The control algorithm is applied for a new type of pendulum (TI Tech pendulum), and the effectiveness and robustness of the proposed control are examined by experiments.

Abstract-Principal mechanisms of passive dynamic walking are studied from the mechanical energy point of view, and novel gait generation and control methods based on passive dynamic walking are proposed. First, a unified property of passive dynamic walking is derived, which shows that the walking system's mechanical energy increases proportionally with respect to the position of the system's center of mass. This yields an interesting indeterminate equation that determines the relation between the system's control torques and its center of mass. By solving this indeterminate equation for the control torque, active dynamic walking on a level can then be realized. In addition, the applications to the robust energy referenced control are discussed. The effectiveness and control performances of the proposed methods have been investigated through numerical simulations.

No abstract

No abstract